Re: [OpenAFS] fileserver and /vicepa
On Jun 3, 2006, at 10:54 AM, vladimir konrad wrote: can openafs fileserver 1.3.81 use other part of the disk then /vicepa? i would need to specify the path where the volumes are to be stored... Well actually it's /viceptwo letters of your choice, but that's not the point... To answer your question... No, the fileserver cannot be configured to look elsewhere (unless you change some code). What's the problem with mounting your storage for AFS to /vicepXX? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] multiple afs client, or have two caches
On May 11, 2006, at 12:19 PM, ted leslie wrote: i'd like to take advantage of the afs caching for added performance, is it possible to have two seperate caches? that is, i know there will be a common set of files used over and over again, that really should remain in RAM cache all the time, but other large files may also be cached and wipe out the other ones that i'd like to always have cached. One way I can fix this , is if i could have two afs clients on the same workstation, one that has access to the afs that have the very popular files, and other that has access to the infrequent, but large files that otherwise would spoil the cache. Can one have two seperate afs clients running on a workstation? Another sol'n might be , to be able to control what is and isn't cachable ? any ideas/thought/sol'n would be appreciated. Just some stupid questions/comments ... to your 'solutions'. If you have files which are always in the cache and never change, why don't you copy them over to that particular machine? (even on a ramdisk, if you need it) You can't get faster than your actual hardware anyway. Who's going to decide what is going to be in what cache? Whoever is doing that, sooner or later it's gonna be wrong ;-) I'm sure it helps a lot more to rethink the location of your data. AFS is a network file system, which means most of the data is on some server. The cache was never thought to hold your complete cell data (that idea got more popular with the decreasing costs of storage). If you don't need it there, don't put it there ... :-) Of course, I don't know what you're doing with that particular client, but a proper configured cell with a good client configuration should really be sufficient for the daily usage of a workstation. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] FreeBSD OpenAFS : when?
On Apr 26, 2006, at 1:05 PM, vladimir konrad wrote: Are there any plans for completing FreeBSD's openafs client functionality in the near future? i found arla working fine (did not do much testing though), i build it from source. http://www.stacken.kth.se/project/arla/ That heavily depends on the version of FreeBSD we're talking about. Arla is not supported on FreeBSD6. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] disaster recovery, help needed.
On Apr 24, 2006, at 12:15 PM, ph rhole oper wrote: On Mon, 24 Apr 2006 09:05:11 +0200, Horst Birthelmer [EMAIL PROTECTED] said: On Apr 24, 2006, at 3:42 AM, ph rhole oper wrote: On Sun, 23 Apr 2006 22:20:19 +0200 (CEST), Chris Huebsch [EMAIL PROTECTED] said: Hello, On Sun, 23 Apr 2006, ph rhole oper wrote: The hd holding the root directory of an afs server has crashed, thus the /var/openafs/ files are lost with it, Was the crashed server only a fileserver or both fileserver and databaseserver? If the latter, was it your only databaseserver? It was a db server too, but there is one more left online.I think that if i install a new OS, mount /vicepa , and start afs services, ubik will synchronize and the new OS's afs will just bring the /vicepa volumes up.Right? The DB server cold make some trouble on resynchronization with an empty database (It's really no big deal) Whenever i need to install a new db server on the network, i just start it with an empty database and it synchronizes with other db servers anyway (is it any other way i could do this anyway?) No. My point was, if it doesn't (which I saw sometimes) it's not part of this problem. ;-) but the fileserver will work, yes. I did that many times in the past. You just have to take care of the KeyFile etc. I'll just copy the KeyFile from another server.. Yes, that's the easiest way, of course. I just wanted to remind you... But the /vicepa partition is ok.Is it possible to recover the data (transfer them to another working openafs server), from the /vicepa/VOL* files? I don't see why a new copy of the fileserver wouldn't do the right thing here. A new copy of the fileserver? Yes, another binary ... from wherever. IIRC it's a linux server. Its FreeBSD actually. If not, you really want to install the same server type (inode or namei). It would be fatal error if you don't. The fileserver of the wrong type can damage your volumes. I'll just use the binary version from the openafs.org site.The previous OS was linux but i don't think i've tampered with the inode/namei option when i built it from source there.. On Linux there is no inode fileserver. You don't have a choice there. :-) Did I get that right? You're trying to attach a partition previously used by a Linux fileserver on a fresh installed FreeBSD fileserver? That kinda changes my statement from above. I don't really know if that won't cause you trouble. You better install the same OS on that machine move all the volumes away and install another OS. From my perspective, it's safer. By the way, if the afs files (KeyFile and /vicepa/*) are generated on a 32bit host, would they be compatible if used on a 64bit host? I mean, can you use them on a new 64bit OS installation?My first guess is no..but i don't know. I don't expect any problems but I don't know for sure either. I never switched from 32 to 64 bits on one machine. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Large Data Storage
On Mar 28, 2006, at 9:18 PM, John Falk wrote: I'm looking for a solution that would provide large amounts of data storage and would be able to grow exponentially. I am a network administrator for a school, like most schools right now budgets are tight and our data storage needs are growing. I am looking to make a giant raid5 out of retired machines. I was looking at the open-afs project to create a network data storage cloud. As machines are retired they would be added to the data storage cloud and all data would be split across several machines like raid5. Is this or can this easily be implemented using open-afs? Short answer to a 'long' question ;-) no. AFS is a 'file system'. This means it _uses_ disks and it doesn't provide some. You can build your 'data cloud' by an AFS cell, but that's one layer above that (what I think you had in mind). AFS can provide a complete distributed filesystem, where clients and users aren't aware of the whole organization of data and servers behind. It comes with a complete set of management tools for the file system, authentication (if you want some, but that's a little too old for some people) authorization and some backup functionality and some nice supplemental file system goodies. It's not a networked RAID substitute nor a high availability data cluster, etc. (I seem to run out of examples of misinterpretations of a 'distributed filesystem' :-) ) I didn't want to scare you off, and I sincerely hope I haven't, I just tried to clarify things. Horst P.S.: There was another project providing networked software disc RAIDs, but I don't remember the name. ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] File too large
On Mar 13, 2006, at 2:53 PM, Juha Jäykkä wrote: kernel) I cannot write a file larger than approximately 2GB in size to my AFS volumes, even from the fileserver itself. The release notes Build it from source and use --enable-largefile-fileserver This is odd, I have 1.3.81 and I'm quite able to write 2 GiB files on the AFS volume. I do not seem to be able to read them, though. Any process trying to access the over 2GB-parts of the files hangs for ever. It cannot even be killed (SIGKILL). Which one is at fault here, server or client? (Everything runs on linux/XFS, except the client cache, which is on ext2.) They could both be the cause of your problem. The large file support has to be in your client as well as your fileserver, if you want to handle large files. ;-) You should be able to mix clients and servers for files 2GB. I also have one 1.4.0 -server. What happens if I put the large file on 1.4.0 and try to access it from 1.3.81 clients? What if I replicate the volume to 1.3.81 fileservers? Shuold I force all fileservers to be of the same version? I don't remember if 1.4.x has large file support enabled by default, since I don't use packages, but if it doesn't, you only get into trouble when you mix in the 'wrong direction'. Which means, you better don't handle large files with a server or client which doesn't have it supported. (kinda obvious, isn't it?) For the replication part, you actually shouldn't be able to release volumes with large files on a server which doesn't support that. On my machines the 'vos release' fails, but I'm not sure there aren't any cases where it appears to be working. I wouldn't trust it anyway... :-) You don't have to 'force' the fileservers to be the same version you just should think about what you're doing when you move or replicate volumes. A mixed environment requires some extra care, but isn't it always like that? :-) Horst___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Migration to a new host failure
On Mar 11, 2006, at 12:58 PM, ph rhole oper wrote: On Sat, 11 Mar 2006 12:17:43 +0100, Horst Birthelmer [EMAIL PROTECTED] said: ... I know that kind of problem and if you're in a position to stop the cell you can work around it easily. This is for adding a new database server to a one server cell. ... 5. stop the old database server Can i just leave the old database server on, so i can have two database servers for redundancy (this at least should be working..) You talked about migrating to a new database and file server. At least that's what I understood. Of course you can have more than one. You can have any number of servers in AFS, only for database servers numbers above a certain amount just don't make sense. ... I just thought that the ubik synchronization would be much more straightforward..im kind of dissapointed if everytime i'll have to do something like this.. It usually is a lot easier. Database servers aren't migrated 'all the time' anyway. :-) That's why I sent that off list with a note that it's a quick and dirty workaround ;-) I actually didn't want that in the archives. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] What filesystem?
On Mar 4, 2006, at 11:46 AM, Christopher D. Clausen wrote: Horst Birthelmer [EMAIL PROTECTED] wrote: On Mar 3, 2006, at 11:30 PM, Volker Lendecke wrote: If you have one or two servers, AFS probably is not worth the hassle. But AFS really pays off when you run out of fingers to count your servers. I see the initial cost in particular when you're new to AFS as relatively high, but in the long run with a lot of servers around the world, you will start to love it. I just wanted to add, that if you reach that size Volker was talking about, you also ran out of options. AFS is the only file system being able to help you in that case. Microsoft's Dfs should scale as well. Of course, it essentially only works on Windows, but if you are mostly a Windows shop, it may be the best way to go. I am a particular fan of the multi-master read-write replication possibilities it offers, something that is not currently possible with OpenAFS. And MS Dfs is suposed to get even better in Windows 2003 R2. My experience with DFS is very limited (I actually didn't have more than a few contacts with it), but is there any possibility for an administrator to move any of the shared data transparently to a new location without all the clients to notice? Is there any possibility to manage the 'shares' at all? I'm a little confused how and why that's a match for AFS. We were talking about a large worldwide installation with possible mountpoints from different cells etc. I quickread some of the information from Microsoft on this topic but didn't find any information about any improvements of DFS beyond that 'name service' for windows fileservers. (well, it's over simplified, I know) Some will argue that the last write wins algorithm it uses isn't a good idea, but for user home directories I think it is the best thing to do and possibly the only solution out there for Windows. Those algorithms weren't point here ... ;-) If there are further questions, I'd suggest discussions on the #openafs IRC channel on the freenode network. I know about the channel. I'm in there, too, most of the time, but not that active ;-) Sorry. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] What filesystem?
On Mar 3, 2006, at 10:11 PM, John Bass wrote: Hi, I hope everyone doesn't mind if I repeat a question. One of the few responses I did get said I should be more complete in asking my question. My boss is trying to choose one distributed file system from NFS and MS-DFS and so on, from my feeling, AFS sounds like the best choice, it is secure and is being used by many large institutes, such as Harvard, MIT, NASA, Stanford, etc. Actually GrandCentral.org says they use it, I cant tell for sure if that means the whole place, or one department, or a couple of people in a lab. A comparison between those three file systems would be highly unfair. First, because we (the people on the list) wouldn't be that objective ;-) and second, because they're very different and neither of them will solve all of your file providing problems in your organization. What features you're defining as important and which you'll never going to need is up to you. Keep in mind - no matter how you and/or your boss will decide - that AFS is a _distributed_ file system. It does start to make sense if you have more file servers. It is highly scalable from one fileserver (which is kinda trivial) to a complete distributed cell with servers placed all over the world. You can choose any level of distribution in between. ... Anybody can see that some of very smart people made and use AFS, but how can an overworked person like me convince my also overworked boss that we should spend the money and time necessary to learn, deploy, train, and support AFS enterprize-wide? This is something for you to decide... It will definitely cost you some money and time. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] What filesystem?
On Mar 3, 2006, at 11:30 PM, Volker Lendecke wrote: On Fri, Mar 03, 2006 at 11:22:55PM +0100, Horst Birthelmer wrote: Anybody can see that some of very smart people made and use AFS, but how can an overworked person like me convince my also overworked boss that we should spend the money and time necessary to learn, deploy, train, and support AFS enterprize-wide? This is something for you to decide... It will definitely cost you some money and time. If you have one or two servers, AFS probably is not worth the hassle. But AFS really pays off when you run out of fingers to count your servers. I see the initial cost in particular when you're new to AFS as relatively high, but in the long run with a lot of servers around the world, you will start to love it. Volker Samba Team ;-) Now that's nice ... I just wanted to add, that if you reach that size Volker was talking about, you also ran out of options. AFS is the only file system being able to help you in that case. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Replication error
On Mar 1, 2006, at 8:05 AM, Juha Jäykkä wrote: Hi! I hit a strange error this morning. I did a little mass-resync of our read-only replicas this morning (it was a bos cron job, ran at 5:30 in the morning with absolutely no users around). Vos release reported success for all volumes, but VolserLog of the fileserver in question says (these are from the RW RO site; there are equivalent lines in the RO fileserver logs as well; each is just 12 seconds later) Wed Mar 1 05:38:57 2006 trans 472 on volume 536870933 is older than 300 seconds Wed Mar 1 05:39:27 2006 trans 472 on volume 536870933 is older than 330 seconds Wed Mar 1 05:39:57 2006 trans 472 on volume 536870933 is older than 360 seconds Wed Mar 1 05:40:27 2006 trans 472 on volume 536870933 is older than 390 seconds Wed Mar 1 05:40:57 2006 trans 472 on volume 536870933 is older than 420 seconds What's going on? I found some mails in the archive with the same log entries but they all had had their vos release return with an error! What concerns me most here is that vos actually told me the volume was released successfully. We are doing regular resyncs of the RO volumes automatically and not getting an error message from something automatic goes awry is quite bad. AFAIK, that's just an informative log entry. It's telling you that the volume release took longer than X seconds. It's nothing wrong with that. Horst___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: OpenAFS + Kerb5
On Feb 16, 2006, at 10:14 AM, Sergio Gelato wrote: * Jeff Blaine [2006-02-15 13:50:52 -0500]: /opt/SUNWspro/bin/cc -O -I/export/home/src/openafs-1.4.1-rc7/include -I. -dy -Bdynamic -I. -c ./config.c make[3]: /opt/SUNWspro/bin/cc: Command not found I don't think this invalidates all the ideas about improving the build scripts in any way, but in the specific case of Sun's unbundled C compiler I read on http://developers.sun.com/prodtech/cc/products/ get.html : Sun Studio 11 software is available as a [...] download, at no charge. So a quick and inexpensive fix might be to install /opt/SUNWspro/ bin/cc . On most UNIX systems you need the 'native' compiler to build the kernel parts. It _won't_ work otherwise. (There were people trying to use gcc on SGI IRIX and IBM AIX.) This is something we know for a long time, and I don't see how that'll change. So, if someone is to redo the compiler recognition make sure all the work is not completely useless. You can't use servers (the userland processes) on HP-UX, IBM AIX, and SGI IRIX systems without having the kernel module loaded, thus you won't gain anything by compiling the userland parts with gcc. You still need the 'native' compilers. The only platform where those checks really matter is maybe Linux. Don't get me wrong, there is room for improvement in the automated recognition magic there, but I don't think it's really worth the effort on all platforms. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: OpenAFS + Kerb5
On Feb 16, 2006, at 4:25 PM, Jim Rees wrote: thus you won't gain anything by compiling the userland parts with gcc. You still need the 'native' compilers. I could imagine some cases where this would be useful. For example, you might be using the pre-compiled package for your platform, but want to recompile some userland tool so you can make a local modification to it. I never thought of that, since for me building OpenAFS was always a sort of 'all or nothing' game ;-) Even though I wouldn't encourage people to distribute different, modified tools also situated inside and built from the OpenAFS tree. But ... you're right. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] VOS commands
On Feb 7, 2006, at 2:58 PM, Juha Jäykkä wrote: ... We have the same setup, but it does not quite work as we supposed it would. We thought that anything accessing the RO version would automatically switch over to the other copy whenever to one it is currently accessing drops out of the network, but it does not seem to do so. Perhaps there is a misconfiguration somewhere? The situation is this: RW version of volume X on volume server A RO versions of X on servers A (on same partition as RW copy) and B Now users' $HOME is, of course, the RW version (/afs/ blablabla/.username) and programs using $HOME will complain when A disappears. Those accessing the undotted version on the lost server should - to my understanding - move over to using server B in a few minutes after A is lost, but they do not. Why is this? The client never switches from an RW volume to the RO copy. If you mounted the home dir with -rw, which you should and I'm sure you did, your client will never switch to any other replica. The replication is for read only data only, remember? ;-) Horst___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Evaluating AFS for in house use, RFCs...
On Feb 3, 2006, at 1:30 AM, ted creedon wrote: Forget Samba. AFS is better and more secure and more hack proof. Now that's a nice statement for an AFS list ;-) BTW, I think the benefit of having that Samba solution is that you don't have to make changes to the clients. I wouldn't be happy to travel to 30+ offices across the world (or wherever) to install the AFS client, even if I'd be totally convinced that it's the 'better' solution. ... and, Samba and AFS are as secure or unsecure as the admins make it. The question is how much work you would have to do, to make it secure, but that's completely beyond the point. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Fwd: OpenAfs on AIX 5.3
On Jan 27, 2006, at 9:59 PM, Enrique Sanchez wrote: Step 5 of the Overview for Installing Server Functionality says: On some system types, install and configure an AFS-modified version of the fsck program. Is there documentation on which systems this needs to be done? It is only true for systems having an inode fileserver, which most of todays AIX systems have not. AFAIK the namei fileserver ist the default one on AIX. If the AFS modifed version of the fsck program is not present in the binary distribution is not present, should I assume it is not required? What filesystem types are supported on AIX 5.3? For the fileserver or the client cache? For the namei fileserver any filesystem should do. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Maximum size /vicepX partition
On Jan 24, 2006, at 2:26 PM, Hans-Werner Paulsen wrote: Hello, the size of a /vicepX partition on an OpenAFS-1.4.0 fileserver seems to be limited to 4TByte. I can create volumes on this fileserver/partition, but when I try to create a file on this volume I get: $ touch xxx touch: creating `xxx': No space left on device The fileserver writes the following lines to the log file: Tue Jan 24 12:40:10 2006 File Server started Tue Jan 24 12:40:10 2006 Tue Jan 24 12:42:53 2006 Partition /vicepd that contains volume 536920876 is full Tue Jan 24 12:44:41 2006 Shutting down file server at Tue Jan 24 12:44:41 2006 Tue Jan 24 12:44:41 2006 Partition /vicepd: -5451620 available 1K blocks (minfree=0), Tue Jan 24 12:44:41 2006 overallocated by 10008912 blocks ... Are you sure, it's the fileserver and not your partitions filesystem that is limited? Can you write data to the partition directly on the fileserver? I mean writing directly on the filesystem of the fileserver (in this case /vicepd). Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] NetRestrict - change existing openafs server to use a single network interface
On Jan 17, 2006, at 1:36 PM, Vladimir Konrad wrote: hello, we have a openafs server (configured before i turned up) with two ehernet network interfaces (one for normal network activity, one for backup access). this is a production server. the operating system is Debian Woody, openafs 1.2.11... the fileserver currently tries to use both network interfaces but i need to make it use only one. i tried to use NetRestrict file (in different locations) but after the restart of the fileserver, nothing changes. i suspect that this has to do something with existing volume entries because vos changeaddr -remove fails with complaint. Now this is rather unusual but possible ;-) Since we don't know what the error is, it's kinda difficult to guess. what is the correct method to remove use of a network interface on openafs fileserver with existing volumes? The correct method is AFAIK vos changeaddr ... which is precisely what you did. You can remove the entries for the volumes of the server in question by using vos delentry than _fix_ your address problem start the fileserver and use vos syncvldb. Of course you can't do this on a production server. Maybe you want to move the volumes to another server, or something. If you don't know where to put VosRestrict, call the fileserver using strace and watch where the fileserver is trying to open the file. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] NetRestrict - change existing openafs server to use a single network interface
On Jan 17, 2006, at 3:56 PM, Vladimir Konrad wrote: Now this is rather unusual but possible ;-) Since we don't know what the error is, it's kinda difficult to guess. # vos changeaddr -remove the-ip Could not remove server the-ip from the VLDB VLDB: volume Id exists in the vldb You can use vos listvl -server ... to list the 'offending' volumes and you should change the address you want to remove from the VLDB first. (usually to the address you want to keep) If you don't know where to put VosRestrict, call the fileserver using strace and watch where the fileserver is trying to open the file. what is VosRestrict? (should not it be NetRestrict?) Yes of course, it should have been NetRestrict. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Windows Client, quota and filemanager - no warning!
On Jan 12, 2006, at 2:13 PM, Lars Schimmer wrote: ... I assumed the filemanagers are using the standard tools windows holds for them. I contacted the authors of totalcommander and asked for some help. It seems they calculate the disk full message from available space :-( Now that one's OK, if you can make sure you're the only process writing to that specific device. How would the guys do that? I think they should rely on what the filesystem is telling them on the particular operation. Even if they do the check for space on every chunk/block they touch, that would be pretty inefficient on any network file system. Thus that one is not AFS's fault ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Last Update not updating?
On Jan 11, 2006, at 5:40 AM, Mike Polek wrote: The vos release won't actually do any volume copying if things are already up-to-date, but it takes a lot longer to issue a vos release on a set of, say, 10 volumes than to check the dates and release the one that has been updated. Any chance of adding a switch to the vos exam command so that if you do something like vos exam my.vol -sync -cell pictage -local you get the old behavior, and without the -sync you get the new behavior? If I issue the 10 vos release commands, it will have to sync anyway, and apparently do a lot more stuff. Since I'm interested in the Last Update field, it would be helpful to have a way to make sure it's accurate. For the times when I don't care, of course the new behavior is preferable. Just a thought. Not a bad idea, since the client application decides which call it uses. As Jeff mentioned, it's inaccurate for about 25 min. So, this matters just for 25 minutes, than your tools would probably work as before. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Last Update not updating?
On Jan 10, 2006, at 3:18 AM, Jeffrey Hutzelman wrote: On Monday, January 09, 2006 04:43:47 PM -0800 Mike Polek [EMAIL PROTECTED] wrote: Good to know. The reason I'm looking at it is that I have a perl script that will release a set of volumes, but only if they need it. It checks the updateDate of the volume and compares it to the updateDate of the cloneID if there is one to figure out if the volume needs to be released. Yup; we have that too. It doesn't notice changes made too close before it runs. That's one of the reasons I'm looking at making 'vos examine' go back to the slower, more accurate results. Jeff, I remember our talk at the 2004 Hackathon about this. The problem was not that the examine took too long. The problem was, that you could cause timeouts on file transfers, if you do a few 'vos examine' calls, due to the disk operations those calls would trigger. I think it's trouble to go back. BTW, 'vos release' won't do any transfer if the RW volumes didn't change. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Failover
On Dec 31, 2005, at 4:12 PM, ed wrote: On Tue, 27 Dec 2005 23:35:42 +0100 Horst Birthelmer [EMAIL PROTECTED] wrote: That's why the replication has to be triggered by the administrator. Can this not be automated via cron for all volumes every 5min for example? It can, of course, but that's still no failover, since you have just one RW copy. If that server goes down, you _can't write_, and besides that, you have to mount the RW volume differently, so your user will have another path to the RW copy. That's definitely not what people call full replication of data. This is documented that way and AFS people never claimed to have this feature (R/W replication of data across fileservers). Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Failover
On Dec 31, 2005, at 4:53 PM, ed wrote: On Sat, 31 Dec 2005 16:43:38 +0100 Horst Birthelmer [EMAIL PROTECTED] wrote: On Dec 31, 2005, at 4:12 PM, ed wrote: Can this not be automated via cron for all volumes every 5min for example? It can, of course, but that's still no failover, since you have just one RW copy. If that server goes down, you _can't write_, and besides that, you have to mount the RW volume differently, so your user will have another path to the RW copy. That's definitely not what people call full replication of data. This is documented that way and AFS people never claimed to have this feature (R/W replication of data across fileservers). Isn't there a backup server in AFS, or am I confusing that with something else, such as volume database backup. Replication has nothing to do with AFS backups and volume database backup is an entirely separate topic. I'm not sure where exactly the confusion is... ;-) Those are three completely different things. If you can elaborate a little more what you meant by backup server what that has to do with replication, maybe we can straighten this. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 29, 2005, at 12:28 PM, Christof Hanke wrote: Jeffrey Altman wrote: Stephan Wiesand wrote: Wouldn't it be an option to not take over the IP address, but just the vice partition? Once failure of the peer is recognized and confirmed (which is a problem, I agree, but not at all AFS-specific): 1) stonith 2) mount the new vice partition and salvage it [ 2a) is there a need to restart the fileserver? ] 3) vos syncvldb You would also require a 4) vos syncserv There must be some thinko in all this, or people would be doing this a lot. What is it I'm overlooking? I believe this scenario will not work because the VLDB entries for all of the volumes that are being mounted by Server B are listed as being on Server A. Since Server A is unreachable, the volume server when performing the vos syncvldb and vos syncserv steps will not be able to verify that the volume is no longer on Server A. Therefore, there is now an irreconcilable conflict that will cause the vos command to write a message to the standard error stream. The vos command never removes volumes from file server machines. The quotes are from the man pages for vos syncvldb and vos sycnserv. I guess, you con circumvent this, by moving the sysid as well. The hotswap may then work as follows: all /vicep* and /usr/afs (tranarc paths assumed) are on an external storage. If/When fileserver A breaks, you shut it down. A standby node B, with the same OS than fileserver A (which is up to this place _no_ fileserver) mounts the stuff on the external storage. Then you startup the fileserver binary on that. I did something like this years ago and it worked, but somehow I still have a bad feeling about doing it ;-) It was using FibreChannel devices, which are very easy to 'move'. It included moving the IP address of the fileserver as well, to avoid the syncvldb and syncserv headache. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 28, 2005, at 1:31 PM, Paul Robins wrote: Hey, I'm looking at implimenting a 4 server system at work, with each server having 2x200 gig hard drives with 3 raided partitions and LVM on the third. I would like to use some form of AFS, however i'm unsure of it's suitability. I would ideally like to somehow network RAID the available storage space. I assume AFS contains no provision for this so if not, would there be any conflict using AFS on an LVM volume on MD raid? I just assume you're planning to do this on Linux. Since there is no other fileserver than a 'namei' on Linux, you can use it on any device (the LVM is usually completely transparent) and with any filesystem you think it suits best. For the rest of the questions, I don't know how you think you can 'network' RAIDs. If you want to use available storage from all of your fileservers, that's what AFS will help you do. You'll have one namespace and from a users perspective you won't have to care about where the storage actually is located. Actually that's what a distributed and even a network file system is all about. If you think of placing the network some layers lower, like in the block device, etc., AFS can't help. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 28, 2005, at 1:55 PM, Paul Robins wrote: I just assume you're planning to do this on Linux. Yes apologies, linux, probably reasonably modern 2.6 kernel Since there is no other fileserver than a 'namei' on Linux, you can use it on any device (the LVM is usually completely transparent) and with any filesystem you think it suits best. Excellent, that makes things so much easier For the rest of the questions, I don't know how you think you can 'network' RAIDs. I was referring to using something like ENBD to create a RAID array across a network That's on what I commented further below. If you put the network somewhere else than in the file system, it's getting more complicated. If you want to use available storage from all of your fileservers, that's what AFS will help you do. You'll have one namespace and from a users perspective you won't have to care about where the storage actually is located. Actually that's what a distributed and even a network file system is all about. Indeed, what i was referring to is having 4 servers and maintaining some form of redundancy, so that the AFS volume (namespace?) could survive a single server failure, whilst having more than the 180 gig currently free on each machine. I kinda assumed it was impossible So, you expect a network interface failure, or machine failure? Since you can have a lot of redundancy with your RAID design. What I meant by 'namespace', is how afs appears on all of your clients. You'll have a (more or less) fix mount point '/afs' for the AFS space. From that point on, everything is 'free'. You can mount volumes from foreign cells, mount volumes multiple times, whatever you want to do. This tree looks the same on all of your clients. The replication of volumes in AFS works only for read-only data. Maybe you read about that, and that's where most people start getting 'design ideas'... ;-) Strictly spoken, you could build that kind of redundancy (something like network based RAID 0) with AFS, but I consider that, abusing the design. If you think of placing the network some layers lower, like in the block device, etc., AFS can't help. Yeah i figured as much, i was just wondering if AFS contained any way of doing the above. Yes and no ... What you plan is doable with AFS, but I'm not sure you'll get an optimal solution by just using AFS. Maybe you have to reconsider, what data should survive a machine failure, etc. and how your disaster recovery would look like. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 28, 2005, at 2:37 PM, Paul Robins wrote: I'll reply in one if that's ok (sorry for the topposting) No problem ... :-) I would expect a disk to be the thing to go to be honest but regardless, i want some system where there is parity data stored on other nodes in this group of machines. Basically RAID5 but networked would be perfect, as that would give me ~ 400 gig of space whilst being able to handle a machine vanishing from the network (the whole machine dies when the disk does, cheap whiteboxes don't you know) :-) I understand the way AFS works with regards to clients seeing /afs, and i did see read only replication, and then running a command to change a read only node(?) into a read write node (i'm sorry if i'm talking crap, i'd read the wiki if i could). This is why i figured perhaps it could be implimented with some sort of networked RAID5, giving me a lot more storage than just RO mirroring one server to the other 3, but whilst still being redundant. You're not talking nonsense at all. It's exactly the kind of statement I was provoking. (Just the term 'node' is from cluster terminology not AFS, but OK. AFS doesn't care about nodes, there are only volumes on fileservers) :-) I don't really recommend that conversion of RO volumes to RWs for backups. If you search the archives for that topic you'll find some of my old statements regarding that. IIRC I said something like: use that only if your disk is gone, like in 'the dog eat your harddisk'. (I didn't look it up ... ;-) ) Well, reading your idea, I think I have to repeat myself. It's doable, especially with something like ENBD, but you have to 'pay' for that with some performance, as far as I can oversee that design. You'll have one fileserver then (or maybe 2) and transfer a lot of data over the network to those fileserver(s), just to transfer that from there to your AFS clients, in the worst case over the same network. It _could_ work, even if it's not really what the designers of AFS had in mind when they built the system. ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 28, 2005, at 3:17 PM, Paul Robins wrote: You're not talking nonsense at all. It's exactly the kind of statement I was provoking. (Just the term 'node' is from cluster terminology not AFS, but OK. AFS doesn't care about nodes, there are only volumes on fileservers) :-) Sorry yes, i originally approached this from a clustering perspective, looking at GFS etc. I don't really recommend that conversion of RO volumes to RWs for backups. If you search the archives for that topic you'll find some of my old statements regarding that. IIRC I said something like: use that only if your disk is gone, like in 'the dog eat your harddisk'. (I didn't look it up ... ;-) ) Well it's more about redundancy, in a pinch I can deal without having to do any sort of RAID5 network rubbish because 180 gig is more than sufficient. Backups are done offsite anyway and performance isn't a major issue. This is used to support a 24/7 callcentre, and the most important thing is that if a disk were to fail and take a machine down, we could bring up a replacement with as close to identical data as is possible. and transfer a lot of data over the network to those fileserver(s), This isn't a major issue, the machines were originally specced to have a second (gigabit) network card connected to their own gigabit switch / VLAN. I suppose I should have just stated my requirements originally and dispensed with all my 'smart ideas' :) I'm still not sure, I could clarify ... If you have just one fileserver, AFS IMHO doesn't really make sense. Make an NFS server of this single fileserver or a samba server if you have windows boxes to talk to. There is no 'distribution' with one fileserver, right? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS with RAID
On Dec 28, 2005, at 3:54 PM, John Hascall wrote: Well that's what i was originally wondering, can AFS provide the ability to replicate the contents of one fileserver to others which can be used redundantly. It appears not at all; I'd still like to use AFS but I do think i'm going to have to go NFS and then some sort of faux raid 1 for redundancy. Has anyone tried to run AFS on top of drdb (http://www.drbd.org/)? It seems to me that AFS would be unaware that it was running with a network distributed redundant block device under the /vicepx filesystems. The namei fileserver is unaware of the file system and device it's running on. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Failover
On Dec 27, 2005, at 8:15 PM, Pierre Ancelot wrote: Hi everyone :) I wanted to know how afs is reacting in the case of a failover cluster... i use keepalived on linux and i wish to have a failover distributed filesystem, like what happens if a replica dies or if the first server dies ? ... And when it will get back up, what should i know about re-synchronisation ? AFS is not a high availability cluster file system. (It's at least not it's primary design target) The replication in AFS is just for read-only data, and is actually designed for being able to distribute the same (read-only) information on more fileservers all over the world. That's why the replication has to be triggered by the administrator. This might _not_ be what you mean by failover. The keepalive is completely irrelevant in this case, since there is no synchronization other than the one triggered by the admin. If you can specify, what you're planning to do, maybe somebody on the list will be able to help you further. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS klog problem
On Dec 14, 2005, at 8:40 AM, lal anil wrote: Hi ... When I try to run .configure with --with-afs-sysname=s390_linux26 and then run make I get the following error #make make build make[1]: Entering directory `/openafs-1.4.0' make finale DEST=/openafs-1.4.0/s390_linux26/dest COMPILE_PART2B=all DESTDIR= make[2]: Entering directory `/openafs-1.4.0' cd src cd config make all make[3]: Entering directory `/openafs-1.4.0/src/config' make[3]: *** No rule to make target `param.s390_linux26.h', needed by `/openafs-1.4.0/include/afs/param.h'. Stop. make[3]: Leaving directory `/openafs-1.4.0/src/config' make[2]: *** [config] Error 2 make[2]: Leaving directory `/openafs-1.4.0' make[1]: *** [build] Error 2 make[1]: Leaving directory `/openafs-1.4.0' make: *** [all] Error 2 Can anyone help me out :)) I remember telling this one before. If you're building on s390 and not the slightly different s390z there is _no_ AFS port. That's why there is no param.s390_linux26.h file. If you want to use AFS on that machine, you have to do the port for s390 with linux 2.6 yourself. In the s390z port there was just a little part of assembler for lwp to be written and the usual bits and pieces, where some structures and such are different from the other Linux versions. It's still Linux after all. ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: 1.4.1-rc2 feedback
On Dec 14, 2005, at 11:26 AM, Thomas Mueller wrote: On Tue, 13 Dec 2005, Russ Allbery wrote: However, we're running into some problems with file server crashes on Solaris 8 with 1.4.0 as well. We're currently working on figuring out what's going on. This Morning I upgraded to OpenAFS-1.4.1-rc2 on Scientific Linux 3.0.5, Kernel 2.4.21-32.0.1.EL.XFSsmp (several machines). ... Let me know if you need access to the fileserver binary or the core dumps. ... Looks like you end up with some 'non-free' packets in the rx_freePacketQueue. I don't actually see why, but the macro RX_FPQ_MARK_USED() detects the used packet, when it's trying to mark the packets as used. Which means, either the flags are set wrong or the packets are really in the wrong queue. At least, that's what I think happens. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Questions about OpenAFS reality
On Dec 13, 2005, at 1:20 PM, Leroy Tennison wrote: I am just learning OpenAFS and am very impressed with what I see so far. As a result I'm now interested in getting a broad overall picture of it and have several questions. A recent post stated that they had about 7000 users in a single cell, I'm wondering what a realistic maximum is (assume 'average' end user file activity - nothing extraordinary). I saw in the archives a refernce to 45k and a statement that UMich had 100,000. Are these (especially the 100,000 for UMich) confirmed numbers? If anyone out there has a larger cell or knows of someone who does could you provide specifics? Right now I'm not so concerned with how many file servers I would have to have or other particulars as I am about what a realistic limit is with a proper configuration. The increasing number of users isn't primarily related to the number of the fileservers (these are very easily addable to a cell) but to the database servers. If you have a very heavy 'loaded' cell the calls to the protection service (ptserver) are worse than teh quantity of any other call in the cell. You can design your volume location so well, that it scales almost optimal for file access, but you still have those calls for permissions from the fileservers ending up at a smaller number of protection servers. (... just some thoughts about what you have to consider in the design of such a large cell) If I'm not terribly mistaken there were cells with more than 130,000 users around some years ago, but I don't know how 'active' those users really were. How does AFS compare in administrative burden compared to the common PC NOSes (NetWare and AD)? Is it more intensive, less or about the same for a given type of user population? A related question is how sensitive is it, do you have to be overly careful in order for things to work correctly? If you're looking for a fancy graphical administration tool like a Windows server will provide. I don't know of any. AFS administration is completely scriptable, so after some time in administering your cell, you'll end up with some nifty scripts or small programs doing almost everything of your daily work. Maybe some automatic work done at night as well :-) How stable and trouble free is the Windows client? (I saw a statement that the Windows server was considered experimental and not being maintained). The later statement unfortunately is true, but the client is used by many people and is well maintained. ;-) I never had any major problems with the client, but I'm not using it that heavily. Is there a Linux GUI for day-to-day administration? Not that I know of. After some time you don't need it. Believe me ... ;-) What is the status of server-side byte-range locking? If this isn't a near-term reality what alternatives do people use (SQL server is obviously a possibility, are there other alternatives for MS Access style databases and other byte-range locking needs)? If you want to use that for databases put into AFS. Don't ... And besides, what do you gain from a database where you share the files? Maybe for Access it's useful, but if we're talking about big databases, you shouldn't do that. What are people doing for printing, particularly Windows printing? They use Samba, if you want to print from windows boxes and your servers are UNIX. Or use CUPS, which is also very portable. But that's a completely different story. What about workstation customization (consistent look and feel from workstation to workstation, workstation restrictions, etc.)? What are the alternatives there? That depends completely on the workstation, doesn't it? This is not really related to AFS. AFS just provides a nice way of having the same files on all the workstations in the same place (nice feature the global namespace, isn't it?) I'm not looking for detailed responses just reasonable approximations of OpenAAFS reality, thank you for any replies. AFS is very well suited for large hetrogeneous environments with a lot of servers lots of users and where scalability is really a 'must have'. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: 1.4.1-rc2 feedback
On Dec 13, 2005, at 10:56 PM, Christopher D. Clausen wrote: ... Is there a way to restart the bosserver process WITHOUT first stopping all other AFS server processes (and thus causing downtime?) I really don't like to wait for a salvage or have to vos move everything to another server. Or, is there any disadvantage to NOT restarting the bosserver process (other than the auto-restart if something else crashes and using the bos command to control things.) And, if something else did crash (likely ptserver,) can it be manually restarted without using bosserver? No. There is no way that I know of for the bosserver to be restarted without stopping the other services as well. That would be a little hard to accomplish, anyway. You can run any of the servers 'by hand'. The problem is you can't manage those services over the network. I have problems with the bosserver on my Solaris 10 box as well ... and I'm running the current CVS head. That's one of my test machines, so I'm not really complaining ;-) I didn't have the time to look deeper into it. Yesterday it looked like the bosserver didn't realize that the salveger or any other process terminated and therefore didn't start any of the other services like fileserver and volserver. The salvager terminated regularly and wrote a 'normal' log but the bosserver was still telling 'salvaging filesystem', and that's about it. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Using DRBD for vice partitions
On Dec 8, 2005, at 11:36 AM, Dr A V Le Blanc wrote: I have been doing some experimenting using DRBD (Distributed Redundant Block Device; see www.drbd.org) for vice partitions. The idea is that two file servers use two disks to provide a single virtual disk using network RAID. If one server goes down for a long period, the disk can be mounted on the other server and its contents restored. This requires: (1) Restarting the server processes, since you can't add or remove vice partitions from a running server. (2) Doing a vos syncv on the added partition, so that the volumes on that partition get attributed to the new server in the data base. Other than this nuisance, it seems to work well enough. Can anyone see any problems in using this approach? No, we've been doing similar things with FiberChannel devices some years ago, too. If you can assure that you never have two fileservers with the same partition active, it should work. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS on SLES9 on s390
On Dec 7, 2005, at 12:09 PM, lal anil wrote: Hi All I am trying to configure AFS clinet on SLES9 on s390 machine. The OS details are as shown below [openafs-1.3.78] # cat /proc/version Linux version 2.6.5-7.97-s390 ([EMAIL PROTECTED]) (gcc version 3.3.3 (SuSE Linux)) #1 SMP Fri Jul 2 14:21:59 UTC 2004 # tar xvf openafs-1.3.78-src.tar # cd openafs-1.3.78/ # cp /boot/symvers-2.6.5-7.97-s390-s390.gz /usr/src/linux/ Module.symvers.gz # cd /usr/src/linux # gunzip Module.symvers.gz # make modules_prepare # ./configure --enable-largefile-fileserver --enable-namei- fileserver --enable-supergroups --enable-fast-restart --enable- transarc-paths --with-linux-kernel-headers=/usr/src/ linux-2.6.5-7.97 --disable-kernel-module --with-afs- sysname=i386_linux26 # ./configure --enable-largefile-fileserver --enable-namei- fileserver --enable-supergroups --enable-fast-restart --enable- transarc-paths --with-linux-kernel-headers=/usr/src/ linux-2.6.5-7.97 --with-afs-sysname=i386_linux26 ... Are you really sure the sysname should be i386 on a s390? I don't think so ;-) BTW, if that's a real S390 and not the S390z, OpenAFS was never tested with kernels 2.6.x, AFAIK. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.4.1-rc2 Mac File Oddities
On Dec 2, 2005, at 7:38 PM, Mike Bydalek wrote: ... -Touch a file.txt on the Mac -Modify with Linux -Open on the Mac the changes are seen. Modify the file. -Open on the Linux, the changes are seen. Modify the file. -Open on the Mac and only the last set of changes were seen. Modify the file and it disppears using ls -al, but not using ls. Is it possible, that your Mac is behind a firewall or somehow not able to get the callbacks? It would completely fit what you're describing. The first open on the Mac gets the file into the cache, and since no callbacks are received you will get that same version from that on until it's discarded from the cache. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS Server - Windows XP Pro
On Nov 27, 2005, at 8:09 PM, Sherman Wilcox wrote: First question is whether or not OpenAFS Server is stable when running on Windows XP Pro. That is, is OpenAFS server for Windows ready for prime time? I ask because I can't seem to get it configured properly using the 'Server Configuration Wizard'. Everything seems to install OK but the wizard crashes very consistently when I run it. The OpenAFS Server for Windows is not stable at all. Jeffrey Altman pointed frequently out on this list, that it's just a bonus. It's actually not really maintained. In some builds the server is working... ... If anyone has suggestions I'd love to hear them. You can use it as an experiment, but please don't do it for production servers. I use the fileserver on Windows 2000, but it's a version prior to the 1.4.0 release. I have regularly problems with the volserver and the salvager (crashing), but it's just for fun and for using some of the disk space for AFS on my Windows box. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: afs vs nfs
On Nov 22, 2005, at 8:52 PM, [EMAIL PROTECTED] wrote: On Tue, 22 Nov 2005 [EMAIL PROTECTED] wrote: I don't know why the AFS community continues to support this convention which breaks location independence. *shrug*. Okay, support was a bad choice of words, why its supported should be pretty obvious. It'd still be nice to see some alternatives more widely discussed. How about some suggestions or even better, code? Maybe I'm too used to the concept, the way it is now, but I don't have any problems with that. I never had. Sorry, but I don't see your point. What exactly is breaking location independence? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] vos release improvement proposals
On Nov 16, 2005, at 5:44 PM, Peter Somogyi wrote: Hi, I'm working on improvements on syncing sites (vos release). Currently I'm implementing rsync into volserver. I'm in a hope it will be a good speedup. And I hope this feature will be accepted into CVS (after when it's ready and works). Please tell me if not. I have further plans for improvements: - use a different, own protocol specially for vos release replacing the current one, branching from vos dump (When syncing, currently the same data is sent across the network which results in a vos dump. I think this dump format is too generic redundant, and we could gain extra speedup and less data to send for small files when the protocol would be fixed, omitting tag-ing.) The dump format is just right... I think your real improvement and valuable contribution should be a new protocol just for the replication release. Which means the only paradigm you should be breaking is the format used by the volume release, _not_ dump and restore. Well there are some 'historical' assumptions made in the dump format, but I don't think that has to be changed right now, even though this kind of ideas are 'floating' around in the community for quite some time now. Wouldn't that create the possibility of determining the difference between two files (not volumes)? If yes, how do you deal with different versions on different sites. Which lets you have n different versions of a file (n being the number of RO volumes)? - profile optimize vos release: if anybody have experience when this vos release gets extraordinarily slow and knows the environmental reason please write. AFAIK it's transmitting all changes to the volume but on file basis, which generates a lot of traffic for small changes to large files. Horst P.S.: just my two coins ;-) trying to make sure you don't change the volume dump and restore parts unnecessarily ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Problems compiling 1.4.0 under AIX 5.3.0
On Nov 10, 2005, at 8:44 PM, Kurt Seiffert wrote: I'm getting errors when trying to compile 1.4.0 under AIX 5.3.0. Here is the last part of the output from running make: case rs_aix53 in *_linux* | *_umlinux* ) cc -O -I/N/u/ seiffert/Libra/openafs-1.4.0/src/config -I. -I. -I/N/u/seiffert/ Libra/openafs-1.4.0/include -I/N/u/seiffert/Libra/openafs-1.4.0/ include/afs -I/N/u/seiffert/Libra/openafs-1.4.0/include/rx -I/N/u/ seiffert/Libra/openafs-1.4.0 -I/N/u/seiffert/Libra/openafs-1.4.0/ src -I/N/u/seiffert/Libra/openafs-1.4.0/src -K -D_NO_PROTO - D_NONSTD_TYPES -D_MBI=void -o compile_et compile_et.o error_table.o -L/N/u/seiffert/Libra/openafs-1.4.0/lib -lafsutil;; * ) cc -O -I/N/u/seiffert/Libra/openafs-1.4.0/src/config -I. -I. -I/N/u/seiffert/Libra/openafs-1.4.0/include -I/N/u/seiffert/Libra/ openafs-1.4.0/include/afs -I/N/u/seiffert/Libra/openafs-1.4.0/ include/rx -I/N/u/seiffert/Libra/openafs-1.4.0 -I/N/u/seiffert/ Libra/openafs-1.4.0/src -I/N/u/seiffert/Libra/openafs-1.4.0/src -K -D_NO_PROTO -D_NONSTD_TYPES -D_MBI=void -o compile_et compile_et.o error_table.o -L/N/u/seiffert/Libra/openafs-1.4.0/lib -lafsutil - ll;; esac cc: unrecognized option `-K' ld: 0711-317 ERROR: Undefined symbol: yyin ld: 0711-317 ERROR: Undefined symbol: yyout ld: 0711-317 ERROR: Undefined symbol: .yylex ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more information. collect2: ld returned 8 exit status make: 1254-004 The error code from the last command is 1. Stop. make: 1254-004 The error code from the last command is 2. Stop. make: 1254-004 The error code from the last command is 2. Stop. make: 1254-004 The error code from the last command is 2. Stop. The version of the gcc I have available to me on this system is: cc -v Reading specs from /opt/freeware/lib/gcc-lib/powerpc-ibm- aix5.2.0.0/3.3.2/specs Configured with: ../configure --with-as=/usr/bin/as --with-ld=/usr/ bin/ld --disable-nls --enable-languages=c,c++ --prefix=/opt/ freeware --enable-threads --disable-shared --enable-version- specific-runtime-libs --host=powerpc-ibm-aix5.2.0.0 Thread model: aix gcc version 3.3.2 I never heard of somebody succeeding to compile OpenAFS on AIX with gcc. I don't think it's possible to build a kernel module without the IBM compiler. Could this be part of the problem? Not the problem you're having right now, but you're headed towards more ... ;-) (with gcc) The errors above are just from yacc. Those symbols are global variables usually defined by yacc. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] openafs 1.4: kaserver crashes every 5 minutes on AIX 5.2
On Oct 27, 2005, at 11:08 AM, Ernst Jeschek wrote: Hello, I've migrated our AFS DB servers (Cell: wu-wien.ac.at) from Transarc to OpenAFS on AIX 5.2. Since one of the new servers is sync site, the kaserver on this machine crashes every 5 minutes, and is restarted by the bosserver. It happens with 1.4.0 RC6 and RC8. It seems to have something to do with lwp (dbx where output below). The two other openafs db servers are running without problems. What have I done wrong? openafs 1.4.0 RC8, configured with --enable-transarc-paths, --enable-debug and --enable-debug-lwp, compiled with vac.C 5.0.2.8. oslevel -r: 5200-07 | root dbx /usr/afs/bin/kaserver /usr/afs/logs/corekaserver | Type 'help' for help. | [using memory image in corekaserver] | reading symbolic information ... | | Illegal instruction (illegal opcode) in . at 0x2002e820 | warning: Unable to access address 0x2002e820 from core | (dbx) where | warning: Unable to access address 0x2002e820 from core | warning: Unable to access address 0x2002e820 from core | warning: Unable to access address 0x2002e81c from core | warning: Unable to access address 0x2002e81c from core | warning: Unable to access address 0x2002e820 from core | warning: Unable to access address 0x2002e820 from core | warning: Unable to access address 0x2002e81c from core | warning: Unable to access address 0x2002e81c from core | warning: Unable to access address 0x2002e820 from core | .() at 0x2002e820 | warning: Unable to access address 0xfcfdff07 from core | FiveMinuteCheckLWP() at 0x100508cc | warning: Unable to access address 0xfcfdfeff from core | warning: Unable to access address 0xfcfdfeff from core | Create_Process_Part2(), line 778 in lwp.c Any help would be highly appreciated. I'm running db servers on AIX 5.2, too, and they're working. I can't think of why a FiveMinuteCheckLWP would cause a crash. What I'm trying to say, is, that's pretty weird, but that's what all bugs are :-) You're sure it works until the first 5 min. check? Does it say anything in the logs during startup? What does udebug say about the quorum during those 5 minutes? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] fileserver monitor
On Oct 20, 2005, at 3:46 PM, Hans-Werner Paulsen wrote: Hello, one of our OpenAFS fileserver is very busy, and I want to check who is doing the top I/O. Neither afsmonitor nor scout can tell me who (client) is using which data (volume). Are there any other tools around to identify the top usage? How about tcpdump ... ? ;-) I don't think you really want to know the used data, do you? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Can't mount AFS
On Oct 14, 2005, at 2:46 AM, Jason Frisch wrote: I have finally managed to get everything installed and BOS running, but now only AFS will not start... --- [EMAIL PROTECTED] root]# service afs start Starting AFS services. afsd: My home cell is 'tsukaeru.net' SScall(137, 28, 17)=0 afsd: Forking rx listener daemon. SScall(137, 28, 48)=0 afsd: Forking rx callback listener. afsd: Forking rxevent daemon. SScall(137, 28, 0)=0 SScall(137, 28, 19)=0 SScall(137, 28, 36)=0 afsd: Calling AFSOP_CACHEINIT: 4000 stat cache entries, 4000 optimum cache files, 32768000 blocks in the cache, flags = 0x1, dcache ent ries 4000 SScall(137, 28, 6)=0 afsd: Sweeping workstation's AFS cache directory. afsd: Using memory cache, not swept SScall(137, 28, 34)=0 SScall(137, 28, 29)=0 SScall(137, 28, 35)=0 afsd: Forking AFS daemon. afsd: Forking Check Server Daemon. afsd: Forking 6 background daemons. SScall(137, 28, 1)=0 SScall(137, 28, 4)=0 SScall(137, 28, 2)=0 SScall(137, 28, 2)=0 SScall(137, 28, 2)=0 SScall(137, 28, 2)=0 SScall(137, 28, 2)=0 SScall(137, 28, 100)=0 afsd: All AFS daemons started. afsd: Forking trunc-cache daemon. afsd: Mounting the AFS root on '/afs', flags: 0. SScall(137, 28, 2)=0 SScall(137, 28, 3)=0 afsd: ]Can't mount AFS on /afs(22) --- When ever I try and stop afs, the server throws a kernel panic and dies. If anybody has any ideas at all it would be most greatly appreciated. It seems I am hitting my head against a brickwall here :-( Your client is started correctly. It must be some error in the configuration. BTW, do you have a directory '/afs'? Do you have a CellServDB and ThisCell file in the place your client expects 'em? I just assume your not using afsdb and dynroot for test purposes. (I might be wrong on this) ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] FindClient: client ... already had conn ..., stolen by client
On Oct 10, 2005, at 2:16 PM, Peter Somogyi wrote: Hi, We get tons (1800/half an hour) of messages in FileLog like this: ... Fri Oct 7 14:52:16 2005 FindClient: client 42f61b10(74c78628) already had conn 42f478e8 (host 6f01a8c0), stolen by client 42f61b10 (74c78628) ... Can it show real errors? If yes, what kind of errors? Whose error: internal bug, or configuration error, or client error? Or it's just an internal trace message to see that connection reusing is working? Shall we ignore them? Yes, you should ignore them since it's the same client, isn't it? It's just a notification. But, just for the record, what platforms are server and client? ... 'cause I wouldn't vow for the AIX client on this. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Disk cache on root filesystem / unclean shutdown
On Sep 15, 2005, at 1:55 PM, Stefaan wrote: I don't think there is any connection between these events. Well, at least I don't think afs can be blamed for frying your libs. I'm asking mainly because I'm packaging this for gentoo, and putting a disk cache on /var/cache/openafs is the default. Wouldn't want to fry other people's rootfs en mass :) Running openafs-latest, I think was 1.3.86 when I started seeing the problems (date was 1st of august). Rootfs is ext3, kernel is linux-2.6.13-gentoo. You ran a fs check? I did after I started loosing my libraries. I've read that you shouldn't run fsck on a /vicepx, but I guess this was only for the namei-fileserver (right?, as opposed to the inode-one where I suppose you can run fsck) and also not for the cache? Is there any up-to-date documentation on this? You shouldn't run fsck on _some_ platforms on an inode fileserver partition. I assume you don't have one of those since it's a lot of pain if possible at all to actually be able to run an inode fileserver on Linux (I don't know anybody doing that on Linux), it's not related to your problem. AFS has nothing to do with your vanished libs. I'm pretty sure of that. The other remarkable thing I saw was: Your cache is on an ext3 filesystem, which is not supposed to be that way, and can cause trouble. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Disk cache on root filesystem / unclean shutdown
On Sep 15, 2005, at 5:13 PM, Stefaan wrote: The other remarkable thing I saw was: Your cache is on an ext3 filesystem, which is not supposed to be that way, and can cause trouble. So what filesystems are supported for the cache then? Only ext2? On Linux, Yes. Since practically everyone has a journalling rootfs, this would again mean you can't put your cache on your rootfs? I don't know what that means to other people ;-) I was just telling... Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Disk cache on root filesystem / unclean shutdown
On Sep 15, 2005, at 5:17 PM, Hendrik Hoeth wrote: Thus spake Horst Birthelmer ([EMAIL PROTECTED]): Your cache is on an ext3 filesystem, which is not supposed to be that way, and can cause trouble. ext3 is fine, as is ext2. reiserfs ist evil ... If Hendrik checked that, maybe it's fine. I had trouble with that on Linux. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Disk cache on root filesystem / unclean shutdown
On Sep 15, 2005, at 6:00 PM, Hendrik Hoeth wrote: Thus spake Horst Birthelmer ([EMAIL PROTECTED]): ext3 is fine, as is ext2. reiserfs ist evil ... If Hendrik checked that, maybe it's fine. I had trouble with that on Linux. What kind of trouble? I'm running my cache on ext3 for a year or so, without problems (linux on x86). So I'd be really interested to hear about the caveats. That was a few years ago. The afsd just got stuck in the cache initialization. I got burned once, and I never tried that since ... ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Having issues with writing to afs via Java process.
On Sep 13, 2005, at 1:43 AM, Pucky Loucks wrote: So Jeffrey, am I to understand that the JAVA API is so old that it probably won't work with the latest code? The reason I ask is that I do need a system that creates volumes and replicates/releases and this is one of the main reasons i was looking at the API. AFAIK, the authentication won't work, but most people are using Kerberos and generate Tokens from the Tickets, thus the approach implemented there won't work anyway. The rest of the API, that has nothing to do with authentication, worked for version 1.2.X. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS client installation on Motorolla MVME6100 PowerPC
On Aug 10, 2005, at 2:32 PM, Konstantin Boyanov wrote: Hi there, I have installed the AFS client from the source on a MVME6100 Power PC, but I can get it working yet. When I type /etc/rc.d/init.d/afs start I get the following error message: Failed to load AFS client, not starting AFS services. This message sounds a lot like your kernel module is not loaded correctly. I actually never looked at the current start scripts since I always have my own. The trivial solution to this is you do a lsmod to check if the kernel module is there. What could be the reason for this? Here are the steps through I went: # ./configure --with-afs-sysname=ppc_linux24 --enable-transarc-paths # make # make dest I copied the entire /motload directory, which appeared after the build in /usr/src/openafs-1.2.8/ppc_linux24/dest/root.client/usr/ vice/etc, AND the asfd file to /usr/vice/etc, and the afs.rc file to /etc/rc.d/init.d/afs # cp -rp /usr/src/openafs-1.2.8/ppc_linux24/dest/root.client/usr/ vice/etc/motload /usr/vice/etc # cp -p /usr/src/openafs-1.2.8/ppc_linux24/dest/root.client/usr/ vice/etc/afs.rc /etc/rc.d/init.d/afs # cp -p /usr/src/openafs-1.2.8/ppc_linux24/dest/root.client/usr/ vice/etc/afsd /usr/vice/etc According to the Manual from openafs.org, that should do to start the AFS client whit the command. # /etc/rc.d/init.d/afs start But it didn't. So can someone give me some ideas what I did wrong (or what I didn't do at all), or maybe how to get more detailed information about the errors ocurring? Here's what I would do first to debug. You locate the kernel module which should be in src/libafs/ MODLOADxxx (it should be called libafssomething). Then after a reboot you load it by hand, calling insmod module file name, now you start the afs daemon by starting src/afsd/afsd and maybe you want to use -debug -verbose parameters, so you'll be able to see what's happening during initialization of afs. That should give you a running client. Of course you need a working network, an afs cell and a proper configuration of your client (CellServDB, ThisCell, cacheconfig and some servers :-) ) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS client installation on Motorolla MVME6100 PowerPC
On Aug 10, 2005, at 5:41 PM, Konstantin Boyanov wrote: Hi again, Ye, the servers should be there for sure :-) There is a network already working, with a running cell on it, so I won't need to set these up :-P Well, the kernel module was indeed in src/libafs/MODLOAD-2.4.25- rm01.mvme6100-SP/libafs-2.4.25-rm01.mvme6100.o (I suppose it should be a .o file. Or?) I restarted the system and then loaded it by typing the command: #insmod libafs-2.4.25-rm01.mvme6100.o with some warnings printed out: # Warning: loading libafs-2.4.25-rm01.mvme6100.o will taint the kernel: no license # Module libafs-2.4.25-rm01.mvme6100 loaded, with warnings Could that cause me troubles later on? No, that's just the usual license talk of the picky Linux kernel ;-) When i tried to start the afs daemon in /src/afsd I got the following output: afsd: My home cell is 'xxx.de' ParseCacheInfoFile: Opening cache info file '/usr/vice/etc/ cacheinfo'... ParseCacheInfoFile: Cache info file successfully parsed: cacheMountDir: '/afs' cacheBaseDir: '/usr/vice/cache' cacheBlocks: 236000 afsd: WARNING: cache probably too small! afsd: malloc() failed for cache file inode table with -25 entries. The calculation of your cacheFiles goes wrong. I have no clue yet, why ... What's wrong with this one? I increased the cache size but the problem persists. Got any ideas? What file system is your cache on? On that embedded system I think you won't have any ext2 partition for afs cache, do you? Can you test the client with memcache? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS client installation on Motorolla MVME6100 PowerPC
On Aug 7, 2005, at 12:38 PM, Konstantin Boyanov wrote: Hi again, First thank you for replying. I can't get your point - first you say that nobody ever compiled the AFS kernel module for that processor (which is also my oppinion after a hard search through the net), and then you advise me to copy the kernel module into the boot image. Am I supposed to load the AFS client as a module?And excuse me for the lame question but how you define a place in the boot image as appropriate? I wasn't sure what _your_ point was. I couldn't distinguish if your problem was in compiling or getting the kernel module onto the target or configuring the system. Sorry for confusing you. BTW, OpenAFS can only be compiled as a kernel module for Linux (not built static into the kernel) and an appropriate place would be any you like on the root filesystem. Sorry for the confusion again. I'm used to transfer the root fs as well together with the boot image onto the target and boot from flash or whatever there is. Stupid wording mistake... You can, of course, have your root fs somewhere else. (NFS, etc.) For client functionality you need the kernel module, the afsd deamon and a proper configuration (this means all configuration files). Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS on Mac OS X 10.4 and wiki...
On Aug 5, 2005, at 2:40 PM, Lars Schimmer wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi! Is there a version of OpenAFS which works on Apple MacOS X 10.4 ? Yes, get the openafs-1_3_82-macos-10_4 branch from cvs. It should compile and work on tiger. (It does for me ;-) ) Chaskiel provided some binaries a while ago, but I don't remember the URL. And the wiki hits an error: https://grand.central.org:444/twiki/bin/view prints out: grand.central.org has received an incorrect or unexpected message. Error code: -12227 Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS client installation on Motorolla MVME6100 PowerPC
On Aug 5, 2005, at 3:22 PM, Konstantin Boyanov wrote: Hi everybody! I'm merely a newbie at all the AFS stuff, so I will ask you for some help. I'm trying to install a AFS client on a Motorola MVME6100 PowePC running Linux Live Red Hat 8.0 but I cannot find the appropriate guide and documentation for it. I've been googling around the net for some info but all I could find was how to install it on normal Linux machines. If you have something concerning this issue in mind - pleas reply. AFAIK, the MVME6100 is an embedded system based on the MPC7457 from Motorola. I think nobody ever compiled the AFS kernel module for that processor. (well, I might be wrong ... ) Since you have a 'normal' Linux kernel, the installation itself (after you have a working kernel module) would be more or less like on any other Linux system, except maybe you don't have the ability to work on the machine as easy as on another workstation. (copy the kernel module into your boot image in an appropriate place, put afsd in there, place all the configuration - ThisCell, CellServDB, cacheinfo - in the right place, and start AFS) I just did that on Intel based embedded systems before. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] access 2 different cells from one workstation
On Aug 4, 2005, at 8:38 PM, Ron Croonenberg wrote: Hello all, I have 2 OpenAFS servers let's call it csc and afs-1 I have a workstation and want to configure the client to be able two access both the afs-servers/cells. (So on the workstation on /afs both csc and afs-1 are mounted and when I look at /afs I would see : drwxrwxrwx4 root root 2048 Sep 6 2001 .csc.depauw.edu drwxrwxrwx4 root root 2048 Sep 6 2001 csc.depauw.edu drwxrwxrwx4 root root 2048 Sep 6 2001 .afs-1.csc.depauw.edu drwxrwxrwx4 root root 2048 Sep 6 2001 afs-1.csc.depauw.edu I tried to change the config files by added another entry in the CellServDB, but then what ? (when I mess with the ThisCell file it only takes the first one) Can it be done ? (On a windows workstation one can access multiple cells) How is it done (on Linux) From the information you provided I'm not quite sure what you're trying to accomplish. If your listing looks like the one above, what do you want more? If not, are you using -dynroot on your clients? (I don't know the name of the switch for a windows client was) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] access 2 different cells from one workstation
On Aug 4, 2005, at 10:00 PM, Ron Croonenberg wrote: I want to mount both cells in /afs. I can only get one of them there. The listing is more what I WANT it to look like Ok, so what you want is simply mount the root.cell volume of the one cell into the root.afs of the one you're the administrator of. If ACLs are set properly you can browse the whole filesystem ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS suggestions
On Jul 29, 2005, at 12:14 AM, [EMAIL PROTECTED] wrote: On Thu, 28 Jul 2005, Horst Birthelmer wrote: Noora Peura did her masters thesis on that topic. (It was actually for arla, but I guess it could help anyway) Maybe your student already knows her work ... ;-) The report is at: http://www.stacken.kth.se/~noora/exjobb/files.html interesting. what i'd like to see is just read-one/write-all used in conjunction with the existing ro-clones. the problem with ro- clones is that you can't be constantly releasing the rw volume to the ro volumes on every write. if you batch up writes you expose yourself to losing data which has been written due to failure of a single fileserver. to provide acceptable (for my applications) data integrity, just write-all to 2 or 3 fileservers and on failure of any one of them, use one which didn't have the write fail and replicate it to the ro-clones. then create a new r/w bucket and keep going. That won't work. You talked about integrity. That's exactly the key word here. What should happen if one of the fileservers comes back with an old version of every file in the volume? What should happen if a client wrote to a fileserver and that one fails? (when we still have a connection to that server and no replication...) How are your replicated volumes reacting to that? Those are just a few very simple scenarios that came to my mind without really thinking much about it. It's a lot _more_ complicated than that and I'm not talking about the changes to the protocol and the servers which aren't compatible any more so that we can't use servers and clients from different versions. for an append-only service writing into volumes the ro-clones are the right model to store the data long-term -- but while the volume is r/w it needs to still be more available than a single fileserver. The guys who did the design of AFS where neither stupid nor naive. They knew what hell it would be to do replication any other way it was done ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS suggestions
On Jul 27, 2005, at 11:27 PM, Chris Huebsch wrote: On Wed, 27 Jul 2005, [EMAIL PROTECTED] wrote: I'm intrigued by the (yet)? We have a student working on the problem. But he is still evaluating alternatives. Noora Peura did her masters thesis on that topic. (It was actually for arla, but I guess it could help anyway) Maybe your student already knows her work ... ;-) The report is at: http://www.stacken.kth.se/~noora/exjobb/files.html Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Definitive list of AFS Limits?
On Jul 21, 2005, at 7:16 AM, Christopher Mason wrote: Hello. ... Minimum file size (overhead): FAQ says 1k file occupies 1k, but doesn't address overhead. That's minimal and heavily depends on you directory structure. But not noticeable at all. I said depends on your structure since you have mount points, ACLs etc. to handle on the fileserver. I never heard anybody complain about that in the past. (whatever that now means to you ...) Maximum file size: 2GB??? (is this still true?) https://lists.openafs.org/pipermail/openafs-info/2002-November/ 007006.html The release fileservers do have this limit. If you're compiling the fileservers yourselves you can switch the large file support on, but that's experimantal AFAIK. Maximum files in a directory: The limit depends on the length of the filenames; if they are all sufficiently short, the limit is around 64K. This one's correct. ;-) Maximum files in a volume: Maximum size of a volume: 4TB https://lists.openafs.org/pipermail/openafs-info/2004-November/ 015530.html This would be approximately accurate as well (AFAIK it's not exactly 4TB but somewhere in that area, never cared about the accurate number, though :-) ) Number of servers for a read/write volume: 1 (by design) YES. Maximum number of servers for a read only volume: Maximum size of a partition: (OS limited? -- linux = 9TB?) OS limited since the partition is on a file system on your fileserver. (Be aware that there needs to be a special compile-time option enabled to support blockdevices larger 2TB.https://lists.openafs.org/pipermail/openafs-info/2005-April/ 017336.html) Can't comment on that since I never had such devices :-) Maximum cache size: My experience is that you want your client cache to be at least as large as the largest files you are using. Performance on files that won't fit entirely in the cache is terrible. ... Setting cache size is a bit of a black art. https://lists.openafs.org/pipermail/openafs-info/2005-April/ 017552.html Yes, cache size and parameters for the clients is more like a religion :-) Total size of largest known AFS installation: Now here i think you won't get any proud and cocky answer since the AFS guys are not running around with the number of TBs around their neck for every body to know. :-) BTW, if you're using AFS it's the same way of handling data whether you handle 2.1GB or 200TB. Can an AFS volume be grown in size? Shrunk in size? While online? They do that all by themselves since a volume is not a physical unit but an administrative one. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] pre-fetch cache
On Jul 19, 2005, at 2:58 PM, Brian May wrote: Jeffrey == Jeffrey Altman [EMAIL PROTECTED] writes: What would be nice is a special volume type that is a combination of the cache and replicated read-only volumes. If the master copy of any file in this volume changes, then all replicated copies of this file are deleted, and the next request is forwarded to the master server. Write requests are forwarded to the master server. That way multiple clients at one site can share the one cache copy. Jeffrey I don't understand how this would help share a cache. Jeffrey The cache data would still have to be read over the Jeffrey network. The replicated copy could be installed on a fast LAN when the master server is on a slower WAN. This would help in the situation where a number of clients at one site need to access data from a remote site. If the cell design is made that way, yes. But don't blame that on AFS. AFS chose by design caching is done on the client side. That's about it ... I need to talk to my client more about their requirements - I suspect in this case it would be better if the servers were distributed, not centralized as was requested. Still, I think the above would be a good feature for applications that require distribution of large files across a large number of clients across a small number of sites. Caching on the clients is good, but it requires repeatedly downloading each file on each computer at a site, which could be expensive if it has to be downloaded from a WAN connection. Do I get you right?? You're trying something like the every cache manager is a server in some way thought? Don't think further. That's not what AFS was ever designed for. Examples of large files could include video files, CAD designs, etc. This is simple if files are only changed at one site (use rsync + http mirrors for example), it becomes more complicated if files can change at any site. AFS has support for the first of you ideas. The second one is not that easy to solve. I don't know of anybody ever achieve to have a stable solution for that. I don't think you want that, if you think of what that means to your server to server traffic. You talked before about servers on WANs which means you don't have that much speed there. If you define site as a client they already can ... :-) but I'm sure, you already got that. Jeffrey Keep in mind that the Windows cache is now persistent. Jeffrey You can cache up to 1.3GB on the client. So you only Jeffrey need to download the data once. If you are planning on Jeffrey reading the same data repeatedly across numerous reboots Jeffrey you might not need to precache as much as you think. The argument was we don't want to load the network during peak periods. I suspect not much thought was put into this. What is this here? Am I correct in my understanding of AFS that any file writes require the entire file be copied back to the master server as soon as the program calls close() on that file? I think my client forgot to consider this. The client somehow wanted read/write access, guaranteed access to latest version, file locking when a file is opened (anywhere), and no files to be transferred except at night. That sounds a lot like the disconnected idea to me. Which is thought of for a long time but not ready yet. Unfortunately, some of these conflicting requirements are simply not possible without adding time travel capabilities (grin) to AFS. I think time travel is beyond the budget of my client... Maybe if you're able to be more specific, we can talk more than just ideas and give more than just maybe and definitely not answers. If your customer has that much of a specific plan about requirements of the system. Maybe they need a customized distributed file system. :-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] pre-fetch cache
On Jul 19, 2005, at 4:42 PM, Ken Aaker wrote: Hi Jim, I browsed through the paper and it looks close to what I was thinking about, so I'd suspect that the general conclusions probably still hold true. If there were iAFS servers introduced, I think they would be delay servers. What I was thinking was a little different, and might be something that I read about in one of the IBM Storage Tank descriptions. I was thinking that the file server would use a callback to the clients to say that the chunk that you want is at this address (of a closeby client cache) with this ID. The chunk could be transferred and the chunk registered with the fileserver when the transfer is done??? Are you sure it wasn't Distributed Stoarge Tank? Since the simple storage tank has nothing to do with real distributed file systems since they don't distribute the files themselves but the calls to the devices. (In this case the storage devices) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Idea: tree of volumes/mounts...
On Jun 15, 2005, at 11:39 AM, Lars Schimmer wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi! It's me again... I've had a idea right before opening this window. We've had only one RW copy of some root nodes of our AFS cell tree, we thought it was not needed to backup them, it's easy to reproduce them if they went off. E.G.: /afs/cg.cs.tu-bs.de/cd is only once available as a RW and under CD are only other volumes, no more data. But the cd volume got lost and now we are a bit in doubt which volume was mounted under cd at all... So the idea: Is there a tool/script/idea which searches the volumes/vldb and prints out a tree of the cell with the names of the volumes behind? E.G.: cg.cs.tu-bs.de-root.cell |-cd-cd ||-game1-cd.game1 ||-game2-cd.game2 |-home-user ||-usera-user.usera ||-userb-user.userb |-dev-arbeit |-work-work ||-usera-work.usera I think I know what you're trying to accomplish here ... but you don't have this kind of information in the vldb. What you end up doing here is not going through your vldb but going through your afs tree and filling a database whith your mount points. You will end up with a list where every volume is mounted in the tree if it is at all. From that on you can interpret that data however you see fit. I think I saw something like that in the scripts from Russ Allbery... maybe he can comment on this. and so on and so on. Made for a easy and quick overview over the cell. Because if you lost some of the root nodes, you are lost within the tree if you don know it inside your brain where the mounts are located at... From my point of view, that's what backups are for ;-) ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: OpenAFS-info digest, Vol 1 #2399 - 2 msgs
On Jun 6, 2005, at 11:07 AM, sabah wrote: Dear All, I have not seen a reply for the MAC OS 10.4. Has any one found the answer? Many thanks, Sabah. 2. AFS SW for Mac OS 10.4 Tiger (Padiyath Sreekumaran) Message: 2 From: Padiyath Sreekumaran [EMAIL PROTECTED] To: openafs-info@openafs.org Date: Tue, 24 May 2005 10:16:17 +0200 Subject: [OpenAFS] AFS SW for Mac OS 10.4 Tiger Hello, Is there an AFS client SW available for Mac OS V10.4(Tiger)? No, AFAIK it's not finished yet. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] vos replication being strange.
On May 6, 2005, at 10:50 AM, Eric Bennett wrote: Followup; Now on the RO server this volume cannot be access at all service RWrite: 536870924 ROnly: 536870925 RClone: 536870925 number of sites - 2 server corvus.umbralservices.com partition /vicepa RW Site -- New release server raven.umbralservices.com partition /vicepa RO Site -- Old release corvus:/afs/umbralservices# cd service corvus:/afs/umbralservices/service# raven:/afs/umbralservices# cd service -bash: cd: service: Connection timed out Still exact same error on trying to vos release manually What sort of replication is there automatically? or are we expected to write crontabs for vos release at regular intervals? There is no automatic replication in AFS. You can write crontabs or cron services for your bosserver. BTW, I don't remember you telling us what platform and version (of AFS) the two servers are. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] vos replication being strange.
On May 6, 2005, at 12:06 PM, Eric Bennett wrote: Both 2.6.8-2-k7 debian standard kernels (debian sarge) a worrying update; corvus:~# vos listvol corvus corvus:~# vos listvol raven Total number of volumes on server raven partition /vicepa: 4 root.afs.readonly 536870916 RO169 K On-line root.cell.readonly536870919 RO 5 K On-line service.readonly 536870925 RO 30430 K On-line user.readonly 536870922 RO269 K On-line Total volumes onLine 4 ; Total volumes offLine 0 ; Total busy 0 No volumes on corvus? Now, I think that's the reason for your error -bash: cd: service: Connection timed out you posted earlier. Your volumes on 'corvus' are either gone or unaccessable by the clients. The first is very easy to check. Take a look into you AFS partition if the partition mounted there is still there. Which means if you see any files in there the physical access to your partition is working. For the second one if you can afford some out time, you can just restart the filserver by using the bos restart command and let those volumes be attached again. If that doens't work just tell us what's happening ;-) The question still remains how you got there... Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Update from 1.2 to 1.3.81 on Debian Sarge (testing)
On Apr 22, 2005, at 3:48 PM, Vladimir Konrad wrote: this strongly implies that the kernel module is not loaded. does lsmod actually say that a module named 'openafs' is loaded? Yes, the module loads and is loaded (checked with lsmod) before starting afsd from command line... Can you start afsd with -verbose -debug? It'll tell you what syscall returned what error code. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] HP-UX file systems on client
On Apr 20, 2005, at 12:24 AM, rogbazan wrote: Hi, i´m installing a client on a HP-UX, i knew that the file system type where /usr/vice and /usr/vice/etc will be has to be (and only) hfs, is that correct? I don't remember anything like that and I'm pretty sure I've done it on some other file system, too. What you refer to, might be the restriction on the AFS cache. Try using memcache, if you're unsure you have the right file system. I'm using memcache and it works. Could i create those dirs on a volume manager FS? I've done that, too, and the machine didn't bite me :-) but again not for the cache, only for the files to sit around. BTW, what version of HP-UX? All my statements are true for HP-UX 11.11 and undefined for the rest :-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] More on bosserver failures / Suse 9x
On Mar 31, 2005, at 3:21 PM, Steve Devine wrote: Original Problem Suse 9.0 running OpenAfs 1.2.13 Every 24~25 days the bosserver and volserver quits running and a reboot is required. This even affects my db server which is not running any fileserver processes. Latest tests: 1. Running one server Suse 9.0 with newer version of OpenAfs 1.3.78 Still failed at 25 day mark. One difference is the volservers kept running. 2. Running one server Suse9.2 with with newer version of OpenAfs 1.3.78 Still failed at 25 day mark. Volservers kept running. So far the only 'fix' for this is to bos setrestart them for a set time every week. I doesn't matter what the load is on these machines the bosserver checks out somewhere around the 24th day and 22 ~24th hour. Sounds a lot like 2 Giga-seconds ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.3.80 crashing AIX 5.1 (Modified by Kurt A. Seiffert)
On Mar 23, 2005, at 2:01 PM, Kurt A. Seiffert wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 I'm trying to do some testing with 1.3.79 and 1.3.80 on an SP node running AIX 5.1. I am just running the client portion of OpenAFS. I can load the kernel modules and can browse the /afs filesystem without authentication. I can then execute klog successfully and get authenticated. But once authenticated, if I try to browse or cd into the /afs filesystem, the machine crashes with DATA STORAGE INTERRUPT errors. Anyone have any ideas on this? If you look into the archives you'll find a patch from Nikke which eliminates that problem you're describing and maybe a few more. It's not committed yet, that's why you have to apply that patch ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS client on 64 bit LPAR running AIX 5.2 64 bit
On Mar 30, 2005, at 2:17 AM, Raghu S wrote: Hi We are planning to bring in some p-670 LPAR nodes to OpenAFS cell. However, OpenAFS client is not available for AIX5.2 (64-bit) on www.openafs.org downloads. I am trying to collect some info from user community. Anybody compiled OpenAFS on p-series 64 bit and AIX 5.2 (64bit) combination? We are open to use AIX5.2 or Redhat AS or Redhat EL whichever works. If you're using just the client and have a choice... take Linux. The client is more tested and more stable. AIX 5.2 is supported only in the developer versions. The AIX 5.2 client isn't that stable yet. If you read the archives you'll get the picture. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Killing defunct afs processes [was: 1.3.80 won't make on Gentoo Linux i386 2.6.11-r3]
On Mar 24, 2005, at 3:34 PM, Kevin wrote: On Wed, 2005-03-23 at 20:52 -0500, Derrick J Brashear wrote: Change inode to ip on that line and the next line. Thanks, Derrick. That fixed it. While I'm here, can someone offer me any pointers on killing defunct (and related) afs processes? I'm still having problems with running 1.3.80 as a server and I frequently end up with defunct afs processes that won't die after being sent a signal 9 with kill -9 or killall -9. The only way I know to do it now is rebooting and that's getting old fast. The processes you were trying to kill are from a crashed _client_ and there is no known way of getting rid of them than rebooting. (At least no way known to me ;-) ) The Server doesn't use any kernel stuff on linux so any process will terminate on a signal 9. Eg. root 27834 0.0 0.0 0 0 ?S07:09 0:00 [afs_rxlistener] root 27836 0.0 0.0 0 0 ?S07:09 0:00 [afs_callback] root 27845 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27847 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27849 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27851 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27853 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27855 0.0 0.0 0 0 ?S07:09 0:00 [afs_cachetrim] root 27936 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27938 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27940 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27942 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27944 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 27946 0.0 0.0 0 0 ?S07:09 0:00 [afs_cachetrim] root 28017 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28019 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28021 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28024 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28026 0.0 0.0 0 0 ?S07:09 0:00 [afs_cachetrim] root 28031 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28098 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28100 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28102 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28105 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28107 0.0 0.0 0 0 ?S07:09 0:00 [afs_cachetrim] root 28112 0.0 0.0 0 0 ?S07:09 0:00 [afs_background] root 28232 0.0 0.0 0 0 ?S07:11 0:00 [afs_background] root 28234 0.0 0.0 0 0 ?S07:11 0:00 [afs_background] root 28236 0.0 0.0 0 0 ?S07:11 0:00 [afs_background] root 28238 0.0 0.0 0 0 ?S07:11 0:00 [afs_background] root 28240 0.0 0.0 0 0 ?S07:11 0:00 [afs_background] root 28242 0.0 0.0 0 0 ?S07:11 0:00 [afs_cachetrim] root 29336 0.0 0.0 1904 768 pts/6S07:45 0:00 /usr/sbin/afsd -fakestat -stat 2800 -dcache 2400 -daemons 5 -volumes 128 -nosettime root 29337 0.0 0.0 0 0 pts/6Z 07:45 0:00 [afsd] defunct root 29338 0.0 0.0 0 0 pts/6Z07:45 0:00 [afsd] defunct root 29339 0.0 0.0 0 0 ?S07:45 0:00 [afs_cbstart] root 29340 0.0 0.0 0 0 pts/6Z 07:45 0:00 [afsd] defunct root 29341 0.0 0.0 0 0 ?S07:45 0:00 [afs_evtstart] These have all been sent signal 9 multiple times with kill -9 PID or killall -9 process_name. TIA for any suggestions -- -Kevin http://www.gnosys.us ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Killing defunct afs processes
On Mar 24, 2005, at 4:11 PM, Kevin wrote: On Thu, 2005-03-24 at 15:41 +0100, Horst Birthelmer wrote: The processes you were trying to kill are from a crashed _client_ and Right. I should have mentioned that I was running 1.3.80 as both server and client as opposed to strictly a client. Perhaps I should try running it strictly as a server and see if it acts better. there is no known way of getting rid of them than rebooting. (At least no way known to me ;-) ) Bummer. Do they interfere with other afs processes? Meaning, if I were to try starting up other instances of these processes with the os in this state, would the defunct ones interfere with the new ones? I've never successfully done that, since it's just a client that crashed... :-) If it's running on a server I won't reboot the server but get to another machine for administration purposes. I said succesfully since it happens that some client gets started again after crashing (by any script ... etc.) and that never helped. So I can't answer this one for sure. man ps mentions these sorts of processes, but doesn't go into much detail about them. Do they consume memory and cpu time and other resources or are they more artifacts than anything else? IOW, must I reboot (for the sake of stability) after being in this state? Well, your client crashed. How can one know what they're actually doing?? :-) For stability, they don't do any harm but no good either. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Unable to move volumes
On Mar 9, 2005, at 5:16 PM, Hans-Gunther Borrmann wrote: Hello I tried to move several volumes from a TRANSARC fileserver to an OpenAFS fileserver and got the following error message for two volumes (shown only for one): Starting transaction on source volume 537085440 ... done Cloning source volume 537085440 ... done Ending the transaction on the source volume 537085440 ... done Starting transaction on the cloned volume 537089023 ... done Creating the destination volume 537085440 ... done Dumping from clone 537089023 on source to volume 537085440 on destination ...Failed to move data for the volume 537085440 VOLSER: Problems encountered in doing the dump ! vos move: operation interrupted, cleanup in progress... clear transaction contexts access VLDB move incomplete - attempt cleanup of target partition - no guarantee cleanup complete - user verify desired result On the destination the VolserLog contains: Wed Mar 9 16:52:11 2005 VAttachVolume: Failed to open /vicepcp/V0537085440.vl (errno 2) ... and you're sure there are no files left from some other operations like your testing of the savager on your system?? because the code in IH_CREATE... is pretty straight forward. (I just assume it's a namei fileserver) I moved a few GiBs onto my AIX fileserver and back the last days and never had any problems. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] no quorum elected
On Mar 2, 2005, at 6:40 AM, Steve Lenti wrote: Trying an initial install on Suse 9.2. Everything works fine up until I have to add the initial users. Kaserver, vlserver, ptserver, buserver, and bosserver -noauth all up and running. Anyone know the problem I am having below? ka create afs initial_password: Verifying, please re-enter initial_password: Creating user afs : [u] no quorum elected, wait one second Creating user afs : [u] no quorum elected, wait one second ... Try synchronizing the time between your servers. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.3.77 under AIX 5.1: unusable
On Feb 7, 2005, at 11:19 AM, Hans-Gunther Borrmann wrote: On Monday 07 February 2005 07:29, Derrick J Brashear wrote: CRASH INFORMATION: CPU 0 CSA 2FF3B400 at time of crash, error code for LEDs: 3000 pvthread+004700 STACK: [05B53F2C]afs_RemoveVCB+C8 (69746572 [??]) [05B4A328]afs_GetDCache+001774 (33CF1720, , , 2FF3B188, Of course, nothing has changed in afs_RemoveVCB in some time. But, looking at afs_vcache.c the aix 5.1 changes from Hartmut chowed up in 1.3.71. I thought you said you needed to go back to 1.3.65 before things worked. Am I mistaken ? Yes you are mistaken. I always resort to to the kernel extensions of Hartmut from his 15.03.04 built. No extensions of 1.3.x worked for me so far. 1.3.65 server worked for me except probably the salvager. I simply forgot to test it because the last three years I never had to salvage any volume. At the moment I am running an 1.3.77 server in production with fileserver and volserver taken from 1.3.65. The salvager from 1.3.77 does not work. It crashes. The salvager from 1.3.65 doesn't work either. It stops with the following error message: I know that it crashes. And I know the area because mine is crashing, too, since I updated my sources from CVS. I just didn't have the time to do any debugging, sorry, but I talked to Derrick about that backtrace and we weren't sure what that could be. Just a little patience would help perhaps. I know about the problem but just haven't had the time to look into it. What bothers me more is the problem with the volserver. I don't remember me doing anything to that code. ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Urgent: Unable to dump volumes
On Jan 26, 2005, at 6:35 PM, Hans-Gunther Borrmann wrote: Hello all, during the migration from TRANSARC fileservers to OpenAFS namei fileservers I ran into major problems. One is now: I cannot dump volumes which reside on the OpenAFS Fileserver [EMAIL PROTECTED]:www]# vos dump www.uniradio.var -file /tmp/dumptest Error in rx_EndCall VOLSER: Problems encountered in doing the dump ! Error in vos dump command. VOLSER: Problems encountered in doing the dump ! The VolserLog shows: Wed Jan 26 18:30:00 2005 1 Volser: DumpFile: no memory If it is what I think it is ... and I definitely have to go test it a little more than that would be a bug related to one I fixed some time ago. The filesize in the AIX stat is 64bit and if you give that value to a malloc you'll get what the higher 32bits of that number are saying (we have a big endian machine) and in 90% of the cases that's 0. ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Urgent: Unable to dump volumes
On Jan 26, 2005, at 9:12 PM, Hans-Gunther Borrmann wrote: On Wednesday 26 January 2005 21:03, Horst Birthelmer wrote: On Jan 26, 2005, at 6:35 PM, Hans-Gunther Borrmann wrote: Hello all, during the migration from TRANSARC fileservers to OpenAFS namei fileservers I ran into major problems. One is now: I cannot dump volumes which reside on the OpenAFS Fileserver [EMAIL PROTECTED]:www]# vos dump www.uniradio.var -file /tmp/dumptest Error in rx_EndCall VOLSER: Problems encountered in doing the dump ! Error in vos dump command. VOLSER: Problems encountered in doing the dump ! The VolserLog shows: Wed Jan 26 18:30:00 2005 1 Volser: DumpFile: no memory If it is what I think it is ... and I definitely have to go test it a little more than that would be a bug related to one I fixed some time ago. The filesize in the AIX stat is 64bit and if you give that value to a malloc you'll get what the higher 32bits of that number are saying (we have a big endian machine) and in 90% of the cases that's 0. ;-) Horst I just installed fileserver and volserver from 1.3.65 which was the version I used for my tests. These binaries work! Then it wasn't the bug I thought of. That one would have been in there even in that version. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS-Client crashes AIX 5.1
On Jan 18, 2005, at 11:33 AM, Hans-Gunther Borrmann wrote: After the installation of this version on my test machine I started the client, got a token and tard my AFS home directory to /dev/null. After some time the system crashed. Crash information follows: SYSTEM STATUS: sysname... AIX nodename.. ibm1 release... 1 version... 5 machine... 00415FAC4C00 nid... 415FAC4C time of crash: Tue Jan 18 10:53:57 2005 age of system: 14 min., 41 sec. xmalloc debug: disabled CRASH INFORMATION: CPU 0 CSA 2FF3B400 at time of crash, error code for LEDs: 3000 pvthread+004700 STACK: [05B53F2C]afs_RemoveVCB+C8 (69746572 [??]) [05B4A328]afs_GetDCache+001774 (33CF1720, , , 2FF3B188, 2FF3B178, 2FF3B180, 0001) [05B8464C]BPrefetch+8C (05BD2D68) [05B84E98]afs_BackgroundDaemon+0002A8 () [05B2B390]afs_syscall_call+000238 (0002, , 2FF22FFC, D0B2, , 6000) [05B2AE94]syscall+A0 (001C, 0002, , 2FF22FFC, D0B2, , 6000) [3A50].sys_call+00 () This is IMHO not the same problem. It doesn't look like a kernel allocation problem to me. BTW, can you start your client with -nosettime. I had some problems with that in the past, and never got the time to look deeper into it. Maybe it doesn't help but it doesn't hurt either. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] where to put NetRestrict?
On Jan 17, 2005, at 12:23 PM, Hagbard Celine wrote: Hello, This may sound silly, but where's the correct location on Debian where to put the NetRestrict file? That's always a good question ... :-) The best solution is stracing the fileserver. You'll see the attempt to open the file and that's how you know for sure. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 11:09 AM, Hans-Gunther Borrmann wrote: On Friday 07 January 2005 21:39, Horst Birthelmer wrote: On Jan 5, 2005, at 9:04 AM, Dr. Elmar Abeln wrote: Hi all ! i tryed to install Open-AFS (1.3.74) on an AIX 5.2 The Compile part worked without an error. But after loading the kernel extensions (export.ext and afs.ext.32) and starting the AFS-Daemon afsd the machine crashes. The Machine crashes right away or after getting a token?? Since you use the afs.ext.32 module it is possible that you run into that old kernel allocation problem. I have compiled 1.3.77 under AIX 5.1 and see the same problem. In my case the machine crashes after getting a token. It seems to work before. I see this problem with all versions of OpenAFS I compiled. My workaround is always to use kernel extensions I got some times ago from Hartmut Reuter. They work. If this is an old problem, is any patch available somwhere? Yes. In CVS for some time now Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 2:23 PM, Hans-Gunther Borrmann wrote: On Tuesday 11 January 2005 11:14, Horst Birthelmer wrote: On Jan 11, 2005, at 11:09 AM, Hans-Gunther Borrmann wrote: I have compiled 1.3.77 under AIX 5.1 and see the same problem. In my case the machine crashes after getting a token. It seems to work before. I see this problem with all versions of OpenAFS I compiled. My workaround is always to use kernel extensions I got some times ago from Hartmut Reuter. They work. If this is an old problem, is any patch available somwhere? Yes. In CVS for some time now Horst Hello, I've tried the 10.1.05 snapshot and the machine still crashes. What I did: I fetched openafs-snap-2005-01-10.tar.gz and stored the contents of the tar to directory snap. cd /../openafs-1.3.77; make clean rsync -av snap/ /../openafs-1.3.77 # possible (??) or is this my fault call of configure: CC=cc ./configure --with-afs-sysname=rs_aix51 \ --enable-namei-fileserver \ --enable-largefile-fileserver \ --enable-tivoli-tsm \ --enable-transarc-paths \ --disable-pam make make dest Any idea? you shouldn't need the sysname if everything is OK... ;-) Now check if there is a passage in your rxkad.h that looks like: #if defined(AFS_AIX52_ENV) #ifdef __XCOFF64__ #define MAXKTCTICKETLEN 12000 /* was 344 */ #else #define MAXKTCTICKETLEN 344 #endif #else #define MAXKTCTICKETLEN 12000 /* was 344 */ #endif it's pretty much at the beginning BTW, this is a 32 bit Kernel on a AIX 5.2 machine?? Right?? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 3:17 PM, Hans-Gunther Borrmann wrote: On Tuesday 11 January 2005 14:30, Horst Birthelmer wrote: you shouldn't need the sysname if everything is OK... ;-) compiles without sysname. Now check if there is a passage in your rxkad.h that looks like: #if defined(AFS_AIX52_ENV) #ifdef __XCOFF64__ #define MAXKTCTICKETLEN 12000 /* was 344 */ #else #define MAXKTCTICKETLEN 344 #endif #else #define MAXKTCTICKETLEN 12000 /* was 344 */ #endif it's pretty much at the beginning rxkad.h contains these lines. BTW, this is a 32 bit Kernel on a AIX 5.2 machine?? Right?? It is a 32 bit kernel on an AIX 5.1 (!) machine. If it crashes you can try to change the lines above to AFS_AIX51_ENV since that would be defined on a AIX 5.2 machine as well. I don't know for sure since I never tested on AIX 5.1... Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 5:35 PM, Hans-Gunther Borrmann wrote: On Tuesday 11 January 2005 15:24, Horst Birthelmer wrote: On Jan 11, 2005, at 3:17 PM, Hans-Gunther Borrmann wrote: On Tuesday 11 January 2005 14:30, Horst Birthelmer wrote: you shouldn't need the sysname if everything is OK... ;-) compiles without sysname. Now check if there is a passage in your rxkad.h that looks like: #if defined(AFS_AIX52_ENV) #ifdef __XCOFF64__ #define MAXKTCTICKETLEN 12000 /* was 344 */ #else #define MAXKTCTICKETLEN 344 #endif #else #define MAXKTCTICKETLEN 12000 /* was 344 */ #endif it's pretty much at the beginning rxkad.h contains these lines. BTW, this is a 32 bit Kernel on a AIX 5.2 machine?? Right?? It is a 32 bit kernel on an AIX 5.1 (!) machine. If it crashes you can try to change the lines above to AFS_AIX51_ENV since that would be defined on a AIX 5.2 machine as well. I don't know for sure since I never tested on AIX 5.1... Horst Still crashes :-(. Now that's getting pretty strange. The more I hear about it the more I'm convinced we have some other kind of bug here not just the ticket length. BTW, I'm using the current cvs version on my machine and it's working. Can you get a backtrace or something from the crash?? Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 4:22 PM, Michael Niksch wrote: I have compiled 1.3.77 under AIX 5.1 and see the same problem. In my case the machine crashes after getting a token. It seems to work before. I am seeing that same problem with 1.3.77 on all AIX 4.3.3, AIX 5.1, and AIX 5.2, both 32 and 64 bit kernel. In contrast to previous versions of 1.3, I can at least load the kernel extensions now. Obtaining a token with 1.3.77 'klog' from a kaserver causes a core dump, and trying to use a token obtained with 1.2.10 'klog' results in a system crash. Yes, I understand that 'klog' and 'kaserver' are considered more or less deprecated, and we are in fact planning to change to a Kerberos 5 setup, but this migration will take us quite some time to complete. Also, we will need continued interoperability with legacy IBM Transarc AFS cells for quite some time to come. So it would be great if 'kaserver' support wasn't silently dropped out of OpenAFS already at this point. I don't think that has happened but maybe somebody else can comment on that... So far, I haven't been able to compile ANY version of OpenAFS client to work on AIX 5. The server code might be less problematic as it doesn't involve kernel extensions, but I am also still running 1.2.11 server binaries compiled on AIX 4.3.3 on my AIX 5.2 servers. The stable tree has never had any support for AIX 5, in any way ever. If you try any, please use the CVS version. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 11, 2005, at 8:10 PM, Douglas E. Engert wrote: Jim Rees wrote: I don't think this is your problem, but this looks wrong to me: if (inSize AFS_LRALLOCSIZ) { inData = osi_AllocLargeSpace(inSize+1); } else { inData = osi_AllocLargeSpace(AFS_LRALLOCSIZ); } That first one should be osi_Alloc, not osi_AllocLargeSpace. Yes that looks like a bug. A few lines later, osi_Free will be used to free the area. It also looks like insize is not greater then AFS_LRALLOCSIZE as osi_AllocLargeSize tests for this and would panic: osi_Panic(osi_AllocLargeSpace: size=%d\n, size); Also I think this code could use a comment. It's a bit confusing that osi_AllocLargeSpace is being used for small allocs, and osi_Alloc is being used for large ones. Looks like osi_AllocLargeSize and osi_FreeLargeSpace will keep a pool of 4K blocks off of freePacketList. Thus any size 4K get a full 4K. If there is really a large packet like aticket with a big MS PAC, then 4K may not be big enough, so osi_Alloc is used directly. I changed that a long time ago just for testing. You can use any kernel allocation there. If you allocate those 12k+ the system will crash. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] crash on AIX 5.2
On Jan 5, 2005, at 9:04 AM, Dr. Elmar Abeln wrote: Hi all ! i tryed to install Open-AFS (1.3.74) on an AIX 5.2 The Compile part worked without an error. But after loading the kernel extensions (export.ext and afs.ext.32) and starting the AFS-Daemon afsd the machine crashes. The Machine crashes right away or after getting a token?? Since you use the afs.ext.32 module it is possible that you run into that old kernel allocation problem. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.3.x memcache performance
On Dec 27, 2004, at 3:48 PM, Andrej Filipcic wrote: Hi, I have checked diskcache/memcache performance over an ADSL link. The link has 4M download and 512k upload bandwidth. Although there are still problems with disk cache, the copy from afs to local disk is good enough. With disk cache the speed is ~400 kbyte/s (tcp speed is 500 kbyte/s). Memcahe speed is much lower, between 120 and 160 kbyte/s, depending on the fileserver. Is there any reason (bug) for such a big difference? On gigabit network, the copy speed with memcache is never larger than 20 mbyte/s, with disk cache it can go up to 70 mbyte/s. Can you provide more information on your measuring and test environment? I suppose you didn't actually measured AFS performance when you got 70 MB/s with diskcache. That's pretty unlikely. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.3.x memcache performance
On Dec 27, 2004, at 5:39 PM, Kris Van Hees wrote: On Mon, Dec 27, 2004 at 05:28:14PM +0100, Andrej Filipcic wrote: On gigabit network, the copy speed with memcache is never larger than 20 mbyte/s, with disk cache it can go up to 70 mbyte/s. Well, it is not actually performance. dd if=/dev/zero to afs space transfer reaches something like that. This is of course a peak number. Half of time data is written to a local disk cache with no network activity, so on average, the transfer would be 30-40 MB/s. AFS read speed is between 10-20 MB/s. New memcache parameters give similar write transfer now (~60 MB/s on average). So basically, you are measuring a combination of how fast you can write to your cache, together with how fast the AFS client can flush the data from the cache to the server. Do you have writebehind turned on or off on your AFS client? If writebehind is allowed, then you only measure writing to the cache + part of the flushing to the server (since flushbehind allows the close() system call to return before the entire file has been flushed to the server). If writebehind is turned on (what I just don't assume) then what do you get?? The numbers won't be of any significance at all. With the sync-on-close on you know how fast your client will be on a normal, unspectacular, daily usage. IMHO that's the average speed you'll have to expect from that system afterwards in production. Did I miss something?? Are we here in a client speed contest?? ;-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 1.3.x memcache performance
On Dec 27, 2004, at 6:05 PM, Kris Van Hees wrote: On Mon, Dec 27, 2004 at 05:52:21PM +0100, Horst Birthelmer wrote: If writebehind is turned on (what I just don't assume) then what do you get?? The numbers won't be of any significance at all. Correct, which is what I meant. If it is turned out, you'll get (probably) very nice throughput numbers that are meaningless. With the sync-on-close on you know how fast your client will be on a normal, unspectacular, daily usage. IMHO that's the average speed you'll have to expect from that system afterwards in production. Correct, which is indeed also what I meant. Oh, then maybe I got you wrong. That's why I appended the statement below. ;-) Did I miss something?? Are we here in a client speed contest?? ;-) Well, I hope we're not since it is pointless, whichever way you look at it. I do care about performance numbers though, since it helps with the analysis that's going on. Yes, but I think we missed answering the real question in the first place :-) Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] SuSE 9.2: anyone?
On Dec 23, 2004, at 4:54 PM, Sensei wrote: Hi! Has anyone got AFS working on suse 9.2 using their afs client? I had to fix a script (it searched for kernel module libafs, actually the one shipped with suse is called kafs) but anyway, afsd isn't starting: plm02:~ # /etc/init.d/afs-client start Starting OpenAFS Client afsd: Error -1 in basic initialization. Adding cell 'phy.bris.ac.uk': error -1 Adding cell 'vn.uniroma3.it': error -1 Adding cell 'dia.uniroma3.it': error -1 afsd: No check server daemon in client. afsd: All AFS daemons started. afsd: Can't mount AFS on /afs(22) fs: Function not implemented failed plm02:~ # and moreover: plm02:~ # /usr/sbin/afsd -verbose -stat 300 -dcache 100 -daemons 2 -volumes 50 -memcache afsd: My home cell is 'dia.uniroma3.it' SScall(137, 28, 17)=-1 afsd: Forking rx listener daemon. SScall(137, 28, 48)=-1 afsd: Forking rx callback listener. SScall(137, 28, 0)=-1 afsd: Forking rxevent daemon. SScall(137, 28, 19)=-1 SScall(137, 28, 36)=-1 afsd: Error -1 in basic initialization. Your kernel module doesn't work ... if it is loaded properly ... You don't get any AFS syscall processed. Call 17 isn't doing that much dangerous stuff so if the module is working it should be able to execute it. Horst ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Arla / OpenAFS
On Dec 22, 2004, at 10:53 AM, EC wrote: Hi, Maybe a stupid question.. will Arla client work with an OpenAFS server ? Definitely How else would you provide the data for the arla clients since there are no arla servers. They're in an experimental stage. BTW I have that configuration running for some time and the guys from Sweden are running their environment that way for almost forever ;-) Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: user doesn't exist ??
On Dec 19, 2004, at 3:25 PM, Garry Glendown wrote: # klog admin Password: Unable to authenticate to AFS because a pioctl failed. Do you run a client on that machine?? Is the kernel module inserted?? What platform is that?? You're using Linux as a server is that the same machine?? Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] fs fails Connection timed out
On Dec 19, 2004, at 12:08 AM, Garry Glendown wrote: Another problem ... following the instructions, I've gotten pretty far ... accessing the afs directory locally works, but accesses through the fs binary fail with connection timed out ... I tried a reboot, everything seems to work fine, but still the same error ... Could you provide a little more information what exactly you've been doing?? Also, I can't seem to unmount the afs directory anymore once the system is running - keep getting a busy, even though there are no locks (at least none visible, even through lsof) ... That can happen sometimes :-) Actually the unmount should stop the client but I saw that kind of behavior on many platforms, too. I don't care much about that. Nobody wants to stop AFS when it's running... :-)) Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] fs fails Connection timed out
On Dec 19, 2004, at 12:32 AM, Garry Glendown wrote: Horst Birthelmer wrote: On Dec 19, 2004, at 12:08 AM, Garry Glendown wrote: Another problem ... following the instructions, I've gotten pretty far ... accessing the afs directory locally works, but accesses through the fs binary fail with connection timed out ... I tried a reboot, everything seems to work fine, but still the same error ... Could you provide a little more information what exactly you've been doing?? well, just following the documentation step by step ... I got as far as: fs setacl /afs system:anyuser rl (from Configuring the Top Levels of the AS Filespace) Other fs-functions will result in the same error ... According to bos status, all relevant processes seem to be fine: Instance kaserver, currently running normally. Instance buserver, currently running normally. Instance ptserver, currently running normally. Instance vlserver, currently running normally. Instance fs, currently running normally. Auxiliary status is: file server running. Instance upserver, currently running normally. I have not set up any clients (except the server itself) You mean you have a client running on the server?? Otherwise fs won't make much sense, would it?? I mean what good is setting the ACL on /afs when that's not your AFS mountpoint. Since you talked about unmounting AFS I presume you have a client. You loaded the kernel module and started afsd, right?? Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Linux 2.6.10-rc2 - still no luck
On Dec 12, 2004, at 9:48 PM, Steve Wray wrote: While we are on the topic, how far away is openafs support in 2.6 kernel? There are people using it (maybe not on 2.6.10). (it sometimes seems really wierd that a project prefixed by 'open' has better support for Microsoft OS's than open source OS's). This is because there are guys on the Microsoft OSes actually doing the needed work and not complaining about the work being done ;-) The 'open' in the name doesn't say anything about the OS, does it?? Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] unable to start afs client
On Nov 30, 2004, at 1:25 PM, vinod kumar boppana wrote: hello all, i am getting problem with afs client on red hat linux 8.0 (kernel version is 2.4.18-14) . when i run 'service afs start ' with server and client option ON in /etc/sysconfig/afs , the command getting hanged. it only starting server and unable to start the client. the output is : afs lost contact with file server ( all multihome ip address down for server). but i can ping the server with its ip address and so its network is working perfectly. i also did the following. 1. i started the server part with service afs start by having only server option ON in /etc/sysconfig/afs file. 2. then i used /usr/vice/etc/afsd -debug -verbose' , it is giving the following output. afsd: My home cell is 'afs.compunet.barc.in' ParseCacheInfoFile: Opening cache info file '/usr/vice/etc/cacheinfo'... ParseCacheInfoFile: Cache info file successfully parsed: cacheMountDir: '/afs' cacheBaseDir: '/usr/vice/cache' cacheBlocks: 2 afsd: 2000 inode_for_V entries at 0x8075078, 8000 bytes SScall(137, 28, 17)=-1 afsd: Forking rx listener daemon. afsd: Forking rx callback listener. afsd: Forking rxevent daemon. SScall(137, 28, 48)=-1 SScall(137, 28, 0)=-1 SScall(137, 28, 36)=-1 Yes, you can't launch even one AFS syscall. This means you either have not loaded the kernel module or your kernel module doesn't work with your kernel for whatever reason. Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] unable to start afs client
On Dec 1, 2004, at 11:38 PM, EC wrote: Hi Horst !!! I have some news : I've taken my PIV kernel and changed it to PIII. Then applied it to the PIII. Magic OpenAFS client works !!! Which means, since everything is built automatically, there AFS client is dependant of some options in the linux kernel (or is at least not happy with some option enabled..). I'll try to check that out ! As far as I recall you were having trouble with the creation and/or access to VolumeItems. You got an error similar to ENOENT. From your information I don't see any connection other than your previous kernel module perhaps wasn't build for the kernel you were using it on but that you should have been noticing before. ;-) Well, anyway ... since you got the same error with memcache, too, it won't be any option of your file system. The only thing I can think of is, you built against another kernel or you had some of the links the Makefile creates made to another kernel sourcetree. The only option which has some influence here is the one about version information in kernel modules. BTW, enjoy your working AFS :-) Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] unable to start afs client
On Nov 30, 2004, at 1:25 PM, vinod kumar boppana wrote: hello all, i am getting problem with afs client on red hat linux 8.0 (kernel version is 2.4.18-14) . when i run 'service afs start ' with server and client option ON in /etc/sysconfig/afs , the command getting hanged. it only starting server and unable to start the client. the output is : afs lost contact with file server ( all multihome ip address down for server). but i can ping the server with its ip address and so its network is working perfectly. i also did the following. 1. i started the server part with service afs start by having only server option ON in /etc/sysconfig/afs file. 2. then i used /usr/vice/etc/afsd -debug -verbose' , it is giving the following output. afsd: My home cell is 'afs.compunet.barc.in' ParseCacheInfoFile: Opening cache info file '/usr/vice/etc/cacheinfo'... ParseCacheInfoFile: Cache info file successfully parsed: cacheMountDir: '/afs' cacheBaseDir: '/usr/vice/cache' cacheBlocks: 2 afsd: 2000 inode_for_V entries at 0x8075078, 8000 bytes SScall(137, 28, 17)=-1 afsd: Forking rx listener daemon. afsd: Forking rx callback listener. afsd: Forking rxevent daemon. SScall(137, 28, 48)=-1 SScall(137, 28, 0)=-1 SScall(137, 28, 36)=-1 Yes, you can't launch even one AFS syscall. This means you either have not loaded the kernel module or your kernel module doesn't work with your kernel for whatever reason. Horst ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info