Re: [OpenAFS] OpenAFS 1.7.8 windows: no error on quota full on copy action
On 4/5/2012 9:08 AM, Jeffrey Altman wrote: > As I said, the AFS redirector is reporting the over quota > condition as part of the response to the CloseHandle() call. > Explorer does not call FlushFile() before CloseHandle() knowing > that CloseHandle() will flush the file to disk. > > I will think about this some more and see if I can come up > with something. The problem is that Windows is storing the > data into the Windows page cache prior to the AFS redirector > being asked to allocate space. Once we are asked, we fail it > but by then it is too late. I found a bug in the code where the partition free space was being specified instead of the available quota. This will be in the next release. signature.asc Description: OpenPGP digital signature
[OpenAFS] Re: Couldn't get CPS for AnyUser, will try again in 30 seconds; code=5376.
On Thu, 5 Apr 2012 14:17:30 -0500 Brett Heroux wrote: > The output from udebug is: This all looks fine. I probably should have asked for udebug on port 7002; that is what looks weird: $ udebug 74.222.253.110 7002 [...] Local db version is 0.134632965 I am sync site forever (1 server) Recovery state 1f Sync site's db version is 0.134632965 0 locked pages, 0 of them for write That is not a normal db version number; maybe your ptdb has been corrupted. Can you read it from local disk using pt_util? If you run 'pt_util -user' as root, it should spit out a list of all users in the database. Does it do that, or does it complain about some error? Do you see anything in PtLog? (I assume this is in /var/log/openafs, or wherever openafs logs are for you) -- Andrew Deason adea...@sinenomine.net ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: Couldn't get CPS for AnyUser, will try again in 30 seconds; code=5376.
The CellServDB on my system is: >devicesoft.org #EFS 74.222.253.110 #east-gateway.devicesoft.org It resides in /etc/openafs and /etc/openafs/server on both the db/fileserver and the other fileserver. The output from udebug is: root@east-gateway:~# udebug east-gateway 7003 Host's addresses are: 74.222.253.110 Host's 74.222.253.110 time is Thu Apr 5 14:13:10 2012 Local time is Thu Apr 5 14:13:12 2012 (time differential 2 secs) Last yes vote for 74.222.253.110 was 0 secs ago (sync site); Last vote started 0 secs ago (at Thu Apr 5 14:13:12 2012) Local db version is 1333587500.2 I am sync site forever (1 server) Recovery state 1f Sync site's db version is 1333587500.2 0 locked pages, 0 of them for write Last time a new db version was labelled was: 65690 secs ago (at Wed Apr 4 19:58:22 2012) Thanks for your help. Brett Heroux On Thu, Apr 5, 2012 at 12:02 PM, Andrew Deason wrote: > On Wed, 04 Apr 2012 20:06:05 -0500 > Brett Heroux wrote: > > > I have one db/fileserver and another fileserver. The db/fileserver is > > east-gateway. This is the udebug output. > > I don't think you can get a quorum error with just one dbserver. What's > in /usr/afs/etc/CellServDB? (or wherever the server-side CellServDB is) > > > root@east-gateway:~# udebug 7003 east-gateway > > udebug: can't resolve port name east-gateway > > As Brandon mentioned, this is backwards. Seeing this with the arguments > the right way around would still be helpful... > > -- > Andrew Deason > adea...@sinenomine.net > > ___ > OpenAFS-info mailing list > OpenAFS-info@openafs.org > https://lists.openafs.org/mailman/listinfo/openafs-info >
Re: [OpenAFS] BSOD - Vista AFS 1.7.0800
You can follow the links on this page to file a bug report http://www.openafs.org/support.html Hugh Caldwell Astor & Sanders EWeb Systems Administrator United States Geological Survey 703-648-6812 (Office) 703-598-3472 (Cellular) hcaldw...@usgs.gov Room 2C123B From: "Finke, Jon E" To: "openafs-info@openafs.org" Date: 04/05/2012 01:45 PM Subject: [OpenAFS] BSOD - Vista AFS 1.7.0800 Sent by: openafs-info-ad...@openafs.org I am running Vista with the most recent release of OpenAFS. If I enter "\\afs\rpi.edu" into an explorer window, I get a blue screen of death. Is this simply an operator error on my part, or is this something that needs to be reported (and if so, how... Problem signature: Problem Event Name:BlueScreen OS Version:6.0.6002.2.2.0.256.1 Locale ID: 1033 Additional information about the problem: BCCode:a BCP1: 0004 BCP2: 0002 BCP3: BCP4: 82C6DD2D OS Version:6_0_6002 Service Pack: 2_0 Product: 256_1 Files that help describe the problem: C:\Windows\Minidump\Mini040512-01.dmp C:\Users\finkej\AppData\Local\Temp\WER-78936-0.sysdata.xml C:\Users\finkej\AppData\Local\Temp\WERE790.tmp.version.txt Jon Finke - Senior Systems Programmer - CMT - RPI 518 276 8185 (voice) - 518 276 2809 (fax) - http://www.rpi.edu/~finkej ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] BSOD - Vista AFS 1.7.0800
I am running Vista with the most recent release of OpenAFS. If I enter "\\afs\rpi.edu" into an explorer window, I get a blue screen of death. Is this simply an operator error on my part, or is this something that needs to be reported (and if so, how... Problem signature: Problem Event Name: BlueScreen OS Version: 6.0.6002.2.2.0.256.1 Locale ID:1033 Additional information about the problem: BCCode: a BCP1: 0004 BCP2: 0002 BCP3: BCP4: 82C6DD2D OS Version: 6_0_6002 Service Pack: 2_0 Product: 256_1 Files that help describe the problem: C:\Windows\Minidump\Mini040512-01.dmp C:\Users\finkej\AppData\Local\Temp\WER-78936-0.sysdata.xml C:\Users\finkej\AppData\Local\Temp\WERE790.tmp.version.txt Jon Finke - Senior Systems Programmer - CMT - RPI 518 276 8185 (voice) - 518 276 2809 (fax) - http://www.rpi.edu/~finkej ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] Re: Ubuntu
On Thu, 5 Apr 2012 17:20:03 +0200 Sergio Gelato wrote: > My reading of the Ubuntu wiki is that for serious bugs like the ones > in 1.6.0pre1 one can apply for a Stable Release Update. "one can" as in, anyone? Or do we need someone with ubuntu/canonical (or just a debian developer)? > What I usually do is take your Debian source packages and backport > them myself to the Ubuntu releases I wish to support I may be misreading this... are you saying this as something you do just at your site, or are you related to the official Ubuntu packages in some way? On Wed, 04 Apr 2012 10:02:28 -0700 Russ Allbery wrote: > This is at least partly my fault, at least sort of. This is quite a stretch. I have a hard time imagining that pulling things from sid at random points and putting them in a 'stable' ubuntu release without oversight works well, but... apparently it does often enough, if they keep doing it. I know we have people generating PPAs that seem to be paying reasonable attention to OpenAFS releases. Is it a ton of work to try to get that packaging or those people involved with the official Ubuntu packages? I don't know much about how Ubuntu works wrt maintainers and such, but as far as I can tell this is going to keep happening unless we have someone close to Ubuntu actively paying attention. -- Andrew Deason adea...@sinenomine.net ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] Re: Couldn't get CPS for AnyUser, will try again in 30 seconds; code=5376.
On Wed, 04 Apr 2012 20:06:05 -0500 Brett Heroux wrote: > I have one db/fileserver and another fileserver. The db/fileserver is > east-gateway. This is the udebug output. I don't think you can get a quorum error with just one dbserver. What's in /usr/afs/etc/CellServDB? (or wherever the server-side CellServDB is) > root@east-gateway:~# udebug 7003 east-gateway > udebug: can't resolve port name east-gateway As Brandon mentioned, this is backwards. Seeing this with the arguments the right way around would still be helpful... -- Andrew Deason adea...@sinenomine.net ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] Re: Ubuntu
* Russ Allbery [2012-04-04 10:02:28 -0700]: > Jeffrey Altman writes: > > On 4/3/2012 10:04 PM, Ken Elkabany wrote: > > >> 1.6.0pre1 which was packaged with Ubuntu 11.10. Should we make it a > >> priority to upgrade? > > > 1.6.0pre1 is not an official OpenAFS release. It was a tagged > > pre-release. Anyone that deploys pre-releases should be prepared to > > upgrade as each new pre-release and the final release are issued. [...] > However, for packages not independently maintained in Ubuntu, Ubuntu pulls > versions from Debian unstable and then just freezes with whatever they > happened to have at the time. This means that they can end up with some > really unfortunate choices in their releases. > > I'm not sure there's any really good way to fix this, given that I don't > track Ubuntu's release cycle (or use it at all myself). My reading of the Ubuntu wiki is that for serious bugs like the ones in 1.6.0pre1 one can apply for a Stable Release Update. The backports mechanism doesn't really cut it in the general case, since the version to be backported must already be present in some other release of Ubuntu. That leaves PPAs (in the sense of third-party repositories of Ubuntu packages). What I usually do is take your Debian source packages and backport them myself to the Ubuntu releases I wish to support (mostly only the LTS ones). This rarely[*] requires more than a simple run through pbuilder (after adding a suitable ~suffix to the package version string) so I'm not sure there is much to be gained by pooling efforts in this area. [*] Just tried it on 1.6.1-1 and found I needed to undo the debhelper V9 update (or upgrade debhelper in my pbuilders). Easy enough with a peek at the git repository. ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS 1.7.8 windows: no error on quota full on copy action
As I said, the AFS redirector is reporting the over quota condition as part of the response to the CloseHandle() call. Explorer does not call FlushFile() before CloseHandle() knowing that CloseHandle() will flush the file to disk. I will think about this some more and see if I can come up with something. The problem is that Windows is storing the data into the Windows page cache prior to the AFS redirector being asked to allocate space. Once we are asked, we fail it but by then it is too late. On 4/5/2012 8:34 AM, Lars Schimmer wrote: > On 2012-04-05 13:15, Jeffrey Altman wrote: >> At the moment the AFS Redirector does not implement the Windows >> Volume Quota Service interface. Therefore, Windows has no >> knowledge of per user quotas. > >> The behavior you are seeing is due to the fact that Windows is >> copying the entire file into the Windows page cache and it is only >> being sent to \\AFS when the file is closed. At that point the AFS >> redirector does report the over quota error but the application >> ignores it. In this respect the Windows client is now more like >> the UNIX client. > >> Adding support for the Volume Quota Service interface is a feature >> that I would like to see implemented. However, doing so is >> non-trivial. It requires mapping AFS PTS IDs to Windows SIDs for >> users that may access the file system after the quota information >> was reported to the system. > > Thanks > > But until that is implemented, the result is to have always > "unreachable" quotas on volumes for users not to loose data on copy > actions? Do I see it right? > > But: isn´t there any simple, quick hack workaround possible for simple > explorer copy action to tell: file did not arrive 100% at server? > > >> Jeffrey Altman > > MfG, > Lars Schimmer > ___ > OpenAFS-info mailing list > OpenAFS-info@openafs.org > https://lists.openafs.org/mailman/listinfo/openafs-info signature.asc Description: OpenPGP digital signature
Re: [OpenAFS] OpenAFS 1.7.8 windows: no error on quota full on copy action
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 2012-04-05 13:15, Jeffrey Altman wrote: > At the moment the AFS Redirector does not implement the Windows > Volume Quota Service interface. Therefore, Windows has no > knowledge of per user quotas. > > The behavior you are seeing is due to the fact that Windows is > copying the entire file into the Windows page cache and it is only > being sent to \\AFS when the file is closed. At that point the AFS > redirector does report the over quota error but the application > ignores it. In this respect the Windows client is now more like > the UNIX client. > > Adding support for the Volume Quota Service interface is a feature > that I would like to see implemented. However, doing so is > non-trivial. It requires mapping AFS PTS IDs to Windows SIDs for > users that may access the file system after the quota information > was reported to the system. Thanks But until that is implemented, the result is to have always "unreachable" quotas on volumes for users not to loose data on copy actions? Do I see it right? But: isn´t there any simple, quick hack workaround possible for simple explorer copy action to tell: file did not arrive 100% at server? > Jeffrey Altman MfG, Lars Schimmer - -- - - TU Graz, Institut für ComputerGraphik & WissensVisualisierung Tel: +43 316 873-5405 E-Mail: l.schim...@cgv.tugraz.at Fax: +43 316 873-5402 PGP-Key-ID: 0x4A9B1723 -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk99kW8ACgkQmWhuE0qbFyNuPQCeMxihNEgEeNf45OLfJg0Xn4FF qoIAn0+2RtGHGryzyw/at/mCnb/yisH+ =+UJe -END PGP SIGNATURE- ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] OpenAFS 1.7.8 windows: no error on quota full on copy action
At the moment the AFS Redirector does not implement the Windows Volume Quota Service interface. Therefore, Windows has no knowledge of per user quotas. The behavior you are seeing is due to the fact that Windows is copying the entire file into the Windows page cache and it is only being sent to \\AFS when the file is closed. At that point the AFS redirector does report the over quota error but the application ignores it. In this respect the Windows client is now more like the UNIX client. Adding support for the Volume Quota Service interface is a feature that I would like to see implemented. However, doing so is non-trivial. It requires mapping AFS PTS IDs to Windows SIDs for users that may access the file system after the quota information was reported to the system. Jeffrey Altman On 4/5/2012 4:15 AM, Lars Schimmer wrote: > Hi! > > Maybe I do miss a point, but if I try to copy files in a directory > with not enough quota, copy action just stops silently and no error is > shown? > And worse: explorer does show the file is copied with full size... > > E.g. I have a directory with 5 MB quota. A OpenAFS 1.7.8 client on > windows 7 32bit with a 100MB OpenAFS cache. > > If I try to copy the 100 MB file into that 5 MB Quota directory via > drag&drop in explorer, the copy window opens and shows progress with > 6-10 MB/sec up to 90% (size of cache, more or less). Now it hangs some > time. After some kind of timeout the copy windows vanishes and the > target explorer window does show the 100MB file with the expected size > of 100 MB as complete copied into target. > No error was shown. > > If I check with another client, OpenAFS did only copy untill quota is > full, the 100MB file is cut down to 5 MB. > > I do not think thats the expected behavior, or? > > At least it looks dangerous, as users think the file is copied for > 100% and delete the source and do see after some time, files are not > copied 100% into OpenAFS > > > While if quota is already full, the start of a new copy into directory > with full quota is not started, shows a correct error. > > > MfG, > Lars Schimmer > ___ > OpenAFS-info mailing list > OpenAFS-info@openafs.org > https://lists.openafs.org/mailman/listinfo/openafs-info signature.asc Description: OpenPGP digital signature
[OpenAFS] OpenAFS 1.7.8 windows: no error on quota full on copy action
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi! Maybe I do miss a point, but if I try to copy files in a directory with not enough quota, copy action just stops silently and no error is shown? And worse: explorer does show the file is copied with full size... E.g. I have a directory with 5 MB quota. A OpenAFS 1.7.8 client on windows 7 32bit with a 100MB OpenAFS cache. If I try to copy the 100 MB file into that 5 MB Quota directory via drag&drop in explorer, the copy window opens and shows progress with 6-10 MB/sec up to 90% (size of cache, more or less). Now it hangs some time. After some kind of timeout the copy windows vanishes and the target explorer window does show the 100MB file with the expected size of 100 MB as complete copied into target. No error was shown. If I check with another client, OpenAFS did only copy untill quota is full, the 100MB file is cut down to 5 MB. I do not think thats the expected behavior, or? At least it looks dangerous, as users think the file is copied for 100% and delete the source and do see after some time, files are not copied 100% into OpenAFS While if quota is already full, the start of a new copy into directory with full quota is not started, shows a correct error. MfG, Lars Schimmer - -- - - TU Graz, Institut für ComputerGraphik & WissensVisualisierung Tel: +43 316 873-5405 E-Mail: l.schim...@cgv.tugraz.at Fax: +43 316 873-5402 PGP-Key-ID: 0x4A9B1723 -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk99VJUACgkQmWhuE0qbFyNFKQCghw043Gl5WVK4EBU0dSPY2SkH eqYAn3xje4G6jcleLVjFT522YSiUWRJn =6vev -END PGP SIGNATURE- ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
