Re: [OpenAFS] Re: twiki vandalized
Joseph H Vilas <[EMAIL PROTECTED]> writes: > Dammit: I removed as much vandalization as I could last week; it looks > like it took two days for about 75 pages to be re-vandalized. Grrr Frustrating, sounds like we have no backups, is that true ? /Jimmy ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS / FreeBSD5.3
Jim Rees <[EMAIL PROTECTED]> writes: > You'll probably want arla if you want to run a client. If you want to run a > server, try it and tell me what happens. I think "su.se" is running fileservers on FreeBSD with success. /Jimmy ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Tokens, Tickets and two cells...
Frank Burkhardt <[EMAIL PROTECTED]> writes: > $ pts creategroup system:[EMAIL PROTECTED] > pts: Badly formed name (group prefix doesn't match owner?) ; unable to create group > system:[EMAIL PROTECTED] > try this: $ pts creategroup system:[EMAIL PROTECTED] -owner system:administrators /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 160Gb disk space
Enric Font <[EMAIL PROTECTED]> writes: > And, what's exactly your disk configuration? One single partition on every server about 500-1400 GB > I mean, > how many volumes have every computer? i few thousands on every server, maybe about 2000-5000 > how many > computers you have, 4 servers. about 200-400 clients > and how many disk space can > contain a single volume? 1 Gb? can ? i dont know. my largest is about 50 GB 160 GB volumes would be possible but also very uncomfortable. > I can't find it nowhere. What recomend me to do with a > disk of 160Gb in one computer? well, one single partition would do fine. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 160Gb disk space
Enric Font <[EMAIL PROTECTED]> writes: > I want to ask what is the max size that can support > AFS. I was reading the mailing list, and i can't be > able to find some thread about it. I have AFS-servers using 1500 GB RAID-arrays on Linux. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Inquiry: Backing up OpenAFS with IBM Tivoli Storage Manager (TSM)
Christopher Odenbach <[EMAIL PROTECTED]> writes: > Hi, > > at the University of Paderborn we are using OpenAFS on linux servers and > Tivoli on a Solaris Box. We would like to backup our AFS space with tivoli, > but unfortunately up to now IBM has only released an IBM AFS client for IBM > AIX. > > I am actually in contact with IBM and have asked whether it would be possible > to release a client for OpenAFS (on linux or solaris). As this does not seem > to be too easy IBM wants to know if anybody else would be interested in such > a connector. We have the same seup Linux AFS-servers and Solaris TSM-server. We are very intressted in a product from IBM that would backup my AFS-cell in a more supported way. Right now we do full and incementel volume-dumps which we archive into TSM, its all homegrown perl. You can have it if you want it ... /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] 2.6 linux kernel support; afs from ramdisk
Dr A V Le Blanc <[EMAIL PROTECTED]> writes: > On Saturday August 7, Christoph Hellwig, a Linux kernel developer, > gave a presentation at Leeds on Linux filesystems. He said that > AFS was a buggy filesystem with a very messy code base, and that > people should stay away from openafs, which was particularly > unreliable. Well, there is one more AFS-implementation. Or is arla also considered a buggy filesystem with a very messy code base ? If so, could he point out some messyness in the codebase ? /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] newbie question
[EMAIL PROTECTED] writes: > hello, > > when faced with a salvager hung for no apparent reason, is it appropriate to > delete the file /usr/afs/local/SALVAGER.fs and restart the bos server? If you just kill your salvager the bosserver will start a new one. are you sure the salvager is hung ? an a big and slow disk it can take many hours to complete salvage. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Another Doubt - LDAP and SMB
"lbramos" <[EMAIL PROTECTED]> writes: > Hi, > > Now that i have afs running, is it possible to "add" LDAP and SMB? > > Or i have to do everything from scratch again? (please say no... :S) > > Where can i look up for this? > > For LDAP do i really need kerberos5? Or can i work with the original version that > comes with afs? > > Please let me know if you need more information. You could specify "add LDAP & SMB to OpenAFS". What do you want to do ? - export AFS via SMB ? - run SMB instead of AFS ? - do you want to put your users in LDAP instead of using a /etc/passwd ? /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] ReiserFS /vicep, general performance with lots of vols in single part?
Nathan Neulinger <[EMAIL PROTECTED]> writes: > We use reiserfs on /vicep on about 17 servers for around 1.4 TB used. > Works great. We actually don't dedicate a partition either, we use the > AlwaysAttach functionality with a regular dir. Makes managing space a > little easier in some cases. Performance was definately better than with > ext3, but not as good as ext2 in some cases. No real concrete > testing/numbers though. Same as we discovered a few years ago. Here are some numbers: http://www.e.kth.se/~jimmy/afsfsperf/afsfsperf.html especially numbers from jade.e.kth.se are intressting for you. These were the tests i made with afsfsperf before we started to our AFS-Cell from Digital UNIX to Linux. We decided to go with reiserfs and had no known problems related to the filsystemtype ever. I do recommend reiserfs for /vicep* . On our Wiki there is an PerformanceWork page: http://grand.central.org/twiki/bin/view/AFSLore/PerformanceWork /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] [Q] Sudden disappearance of files in volume.
"S.J.Chun" <[EMAIL PROTECTED]> writes: > Hi, > > We have used AFS server clusters(20 servers and 28 clients). But we have encountered > very > strange problem: files in one volume suddenly does not show up. vos examine shows it > takes > 40GB but there's no content. So I did salvage on this volume and found that the > volume now > has 0K of size. What is the reason/cause of this problem? Thanks in advance. Is the volume completly emply ? Is the volume replicated ? /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Cell/Realm Name and DNS
Stefan Nobis <[EMAIL PROTECTED]> writes: > Hi. > > How important is it to choose a domain name as realm and cell name > for Kerberos/OpenAFS? What are the drawbacks if something other > than domain names are choosen? On my site we do not use CellServDB since we belive that AFSDB-rr in DNS is much better. So we do only map up cells in /afs which have a valid AFSDB-rr in DNS, this is an at least 10 year old standard. So if you choose an cellname where you will not be able to add an AFSDB-rr in DNS lets say "idontlikedots", we will never map your cell in our /afs which prevent our users from accessing your cell. So if you ever want to add AFSDB-rr in DNS ,which i recommend, i would pick a cellname that in "in" your domain. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Extremely poor write performance.
"Neulinger, Nathan" <[EMAIL PROTECTED]> writes: > I've built it, and got interesting numbers... How dependent on other > parts of arla is that afsfsperf program well, it uses at least rx and some other arlalib-stuff ... > - it might be worth seeing if it > can be broken out, I dont think its worth it. I dont have the time. > as you have to go and also get kth-krb to get > afsfsperf built. That sucks. It should build with mit-krb . > Alternatively, might be worth providing a set of afsfsperf binaries on > the openafs or grand.central.org sites. Good idea, who fixes the binaries ? How about try to link them static and put them into a tarball ? We could use versionnumber of the arla-version we used to build it. > Then it becomes much more > reasonable to tell someone to "go run afsfsperf". I know that saying "go run afsfsperf" is not reasonable at all. But i did succeded, since we are already discussing where we should put an binary. :-) BTW, where are you numbers ? i could include them in my ugly www-page if you give me all info mentioned there. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS Poll Raw Results
Patrick Boettcher <[EMAIL PROTECTED]> writes: > Hi list, > > on http://turku.wi-bw.tfh-wildau.de/~pboettch/openafs the raw results > of my poll are published. > > If you have any suggestions to me, which kind of statistic is still > neccessary, please let me know. There are at least 250 active and public AFS-cells on the internet. Thats what my afscrawler found out a year ago. And there were about 60.000 AFS-clients. The biggest site was cern.ch in clients and data ammount. I have asked all 60.000 clients what version they were running, we presented that results on The AFS Workshop Dec 2002 in Monterrey. However i have some suggestions to the form. 1) We run arla, openafs and Transarc AFS, i want to choose all of them. 2) I run almost all available versions of clients, can i specify a list ? /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS volumes filesystem
Hartmut Reuter <[EMAIL PROTECTED]> writes: > For the fileserver (with NAMEI-interface which is obligatory for > Linux) you may take whatever you want. We are using reiserfs, other > people ext3. ext2 has the disadvantage of the slow fsck if for some > reason your system should crash. Have you (or somebudy else) tested performance using different filesystems om the same machine? /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AIX 5.1
"Mark Campbell" <[EMAIL PROTECTED]> writes: > Is AIX 5.1 currently supported? No. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] slave volumes
Jimmy Engelbrecht <[EMAIL PROTECTED]> writes: > If the wider connection has better pingtime than the narrow one, it will > suck. Correction: If the wider connection has worse pingtimes than the narrow one, it will suck. ^ /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] slave volumes
"klaas hagemann" <[EMAIL PROTECTED]> writes: > as far as i understood afs, i can create severeal read-only replicas of a > volume. How is the decision made, which of these replicas is taken? In the case of openafs/transarc: Your client will pick one of the sites by random, sites with a lower "serverpref" are more likely to be selected. the serverpref is today calculated by pingtime and giving some bonus if your are located on the same Class-C network. Your client may keep accessing this replica untill it goes away. (at least true for Transarc) In the case of arla: serverpref will be calculated by ping time and nothing else. The client will look at "serverpref" for EACH AFS-operation. Which gives true loadbalancing for sites with almost equal pingtimes. > We have a distributes WAN and want to avoid that the Client takes a volume > which is behind a very thin connection. For availability reasons all the > replicas should stay available but in an ideal situation only the volume > with the widest or fastest connection should be taken first. Has the faster connection also better pingtime ? In this case youre lucky. If the wider connection has better pingtime than the narrow one, it will suck. > Can someone explain me the algorithm used by OpenAFS? There is no general algorithm, algorithm may be slighly different dependent on the version you are using. Arla 0.36 (or at least 0.37) will be using an completly different and more adaptive algorithm. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Alternative AFS-Implementations
FBO <[EMAIL PROTECTED]> writes: > Hi AFS-Fans, Hi! > Does anyone have experience with "alternative AFS-Implementations" > (different from OpenAFS)? > What about ARLA or MS-AFS? What are the differences, advantages, > disadvantages, critical bugs,... We have been running running arla in production over 2 years. We have now about 140 Linux-boxes running arla. * arla does not die when you do "ls -al /afs" * arla never crashes the kernel, if it crashes only the userdeamon dies, which can just be restarted. * we have expierienced better performance on some operations with arla on Linux 2.4 than OpenAFS 1.2.2 * arla's utilities (for ex. vos) does not provide all features you want, we use sometimes the openafs-utilities (which are compatible) on hosts that run arla. * arla does not crash or hang the mashine when you disconnect it from the network. I do not recommend openafs on laptops. * arla does not provide a userful filerserver-environment yet. We run OpenAFS. * arla 0.35 does caching on whole files. I do not recommend arla for Solaris yet. On Solaris we run Openafs. Arla works fine on *BSD, Linux, and almost fine on Tru64. > Performance and stability ist most interesting. > What are the best arguments for and ( I hope they don't exist :-) ) > against OpenAFS? Compared to what ? NFS ? > Are there any Documents available, comparing different AFS-Implementations? i dont think so. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] vos command error while creating first AFS volume on Tru64 5.1A
Jean-Pierre Vorlet <[EMAIL PROTECTED]> writes: > While trying to install the first AFS server on Tru64 5.1A operating > system using OpenAFS version 1.2.3, it was impossible to create the > first AFS volume, root.afs ! > > I have downloaded alpha_dux50.tar.gz at > http://www.openafs.org/release/latest.html and followed exactly the > instructions found at http://www.openafs.org/doc/index.htm. The server should/might be compatible with Tru64 5.1, however i do recommend to compile your own version. The client is NOT compatible with 5.1 or 5.1a! For the client you have to use the arla-client instead, the openafs-client does not work at all. However i have never tested running an arla-client on the same mchine as an openafs-server, but it should work. You could also buy afs from transarc, they have support for your OS. > Everything was Ok until I use the following command to create the > first AFS volume: > ./vos create siuf.unifr.ch /vicepa root.afs \ > -cell unifr.cht -noauth > > This command always return the following error message: > > Failed to create the volume root.afs 536870957 > : I/O error > Error in vos create command. > : I/O error What does the serverlog's says ? is the kernelmodule loaded ? # sysconfig -s | grep afs did you give any special configure-flags ? could you exec the following command also: # vos parti > I have tried to use UFS and AdvFS file system, but same result! > Any help is appreciated. you want to use UFS, dont use AdvFS, dont use AdvFS for the cache on client's either. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Apache WWW Logs to AFS
Blake Atkins <[EMAIL PROTECTED]> writes: > All, > > I have a question about using an AFS server to collect apache logs. If the > logs grow larger than the client cache, what will happen? How often will the > logs be committed to the AFS server? If the webservers or the AFS server > crashes, will data be more at risk to loss than if they were stored locally? > Does anyone have any experience with this? Do not ever write your logfiles into direct to AFS. The data will not be written untill you do fsync() or close(), depending on clientimplementation. We do the following: 1) Log to local disk. 2) Rotate them into afs with cron Works fine for us. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] porting to a new architecture?
Jacob Gorm Hansen <[EMAIL PROTECTED]> writes: > I would suggest looking into the Arla AFS client, since it seems more > self-contained and has already been ported to lots of platforms. The arla-client does already work on an ipaq running linux. Which means, it has already been ported. /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Vos create failure - I/O error
Thomas Vincent <[EMAIL PROTECTED]> writes: > HI Folks, > Any help here would be much appreciated. Creating a AFS cell to test with > here at Apple is harder then it looks. > Thanks in advance, > > #Vos create talker.apple.com /vicepa root.afs -cell cerb.apple.com -noauth > Failed to create the volume root.afs 536870912 > : I/O error > Error in vos create command > : I/O error > > > Sun OS 5.7 Generic_106541-04 sparc SUNW,Ultra-5_10 The only i can say is that i had the same problem yesterday when tested it on Tru64 5.1, the VolserLog shows the following: bash# cat /usr/afs/logs/VolserLog Wed Feb 21 16:56:45 2001 Starting AFS Volserver 2.0 (/usr/afs/bin/volserver) Wed Feb 21 16:56:48 2001 VCreateVolume: partition /vicepa is not in service. Wed Feb 21 16:56:48 2001 1 Volser: CreateVolume: Unable to create the volume; aborted, error code 103 Wed Feb 21 16:56:48 2001 : Value too large to be stored in data type. The tested version was: Openafs-2001-02-15 It didn't happend me on Digital UNIX 4.0d, but i dont remember if i tested the same version. I will look on it as soon as possible. bash# cat /usr/afs/logs/VolserLog /Jimmy ___ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo.cgi/openafs-info