Message boards :
Number crunching :
Why is my host trashing workunits?
Message board moderation
Author | Message |
---|---|
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
https://setiathome.berkeley.edu/results.php?hostid=2103893 I'm not seeing anything in the returned results. This started happening after I upgraded from XP to Vista. The tasks run for 12-14 seconds then error out. Two or three tasks seem to have completed correctly after the OS upgrade, suggesting the problem lies elsewhere. I've already tried resetting the project, uninstalling and re-installing BOINC, and uninstalling and reverting to an older version of BOINC (7.6.33 vs 7.8.3). Thoughts? |
Dr Who Fan Send message Joined: 8 Jan 01 Posts: 3214 Credit: 715,342 RAC: 4 |
Looking at Task 6555596419 Stderr output <core_client_version>7.6.33</core_client_version> Wondering if it is a read/write permission issue with Vista's UAC (User Account Control) denying the read and/or write to the slots folder. UAC is a p.i.t.a. Best to disable it unless you absolutely need it. https://www.symantec.com/connect/articles/how-does-vistas-uac-feature-affect-applications https://www.petri.com/disable_uac_in_windows_vista |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Look for a AV program that hasn't been setup to exclude scanning of the BOINC main and data directories. Reboot the computer. If the problem persists. Reinstall BOINC. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
Looking at Task 6555596419 As a Windows SysAdmin I'm very familiar with what UAC is, and as a fan of UAC (and why I like Vista more than XP), I'd rather keep it enabled. Though you do bring up a good point and I will try running with elevation. Though if some files have completed successfully, wouldn't UAC not be part of the equation? Additionally, since C:\ProgramData\BOINC is not protected by UAC (C:\, C:\Program Files, C:\Program Files (x86), and HKEY_LOCAL_MACHINE are all protected locations), I would think this shouldn't be an issue. Though I will still try anyway. [Edit] I just checked the NTFS permissions on C:\ProgramData\BOINC and C:\ProgramData\BOINC\Slots and the local USERS group have full control, meaning that even without running elevated, standard users have full write access as appropriate. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
Look for a AV program that hasn't been setup to exclude scanning of the BOINC main and data directories. I'm not running AV on this machine. All it does is crunch and I'm behind a very strong hardware firewall. I've rebooted a few times and have re-installed BOINC. Thanks for the input though! |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
I've already tried resetting the project, uninstalling and re-installing BOINC, and uninstalling and reverting to an older version of BOINC (7.6.33 vs 7.8.3). Thoughts? 2GB isn't a lot of RAM, I expect some of that is boing used for the video card/on board video? The Seti CPU applications don't use that much RAM, but combined with only 1 CPU core & very limited system RAM it could be a case of resource issues? I had 4GB of RAM on my 32bit Vista system on a C2D and that was plenty for general usage. Then I put some 2 video cards on it with 2GB of RAM each. Went from having 3.5GB of available system RAM down to about 2GB and the system really was only good for Seti, couldn't do much else on it at the same time. Grant Darwin NT |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
2GB is all that this laptop motherboard will handle, and I just upgraded from 1GB. ;) The onboard video is set at 64MB and Task Manager reports a total of 2014MB of RAM, with 697MB in use and 1.5GB free. As an update, I checked the laptop this morning and it seems to be crunching fine since I explicitly launched BOINC as Administrator. This clue tells me that BOINC is either not playing well in the standard user context or another non-user account is being used to access it's own ProgramData directory. I am not running as a service install and BOINC's startup log shows I'm running under the logged in user's context. The NTFS permissions on the laptop match what I have on my main workstation for all BOINC directories. My main workstation is on a 2008R2 domain functional level domain, while the laptop is on a 2003R2 domain functional level domain. I seem to recall that on my main domain I added the boinc_master and boinc_project accounts to the Domain Users group, but I don't think I did that on the 2003R2 domain. I'll have to investigate further after work when I have time. |
rob smith Send message Joined: 7 Mar 03 Posts: 22204 Credit: 416,307,556 RAC: 380 |
ProgramData is supposed to be user independent - however I did have one "event" when a sub-folder got its ownership messed about with, and like you I could only write using the owner account and administrator account. I had to reset the ownership and access so "everyone" could read and write. It is worth noting that re-installing BOINC does not remove and re-install the data directory, you have to do that yourself, and depending on the rights of the creating user you might end up with only the administrator being able to do anything with the folder (I think that's what I did wrong - but not with BOINC) Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Installing (or re-installing) BOINC in Service Mode does reset the permissions on the data directory, but installing BOINC in user mode doesn't. I suspect that switching from service to user might be meant to reset them, but that probably hasn't been tested to destruction. I have an ancient host - 4060784 - which I found exhibiting similar symptoms a few days ago, but din't have time to investigate. I'll maybe take a look as I continue my 'patch |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
ProgramData is supposed to be user independent - however I did have one "event" when a sub-folder got its ownership messed about with, and like you I could only write using the owner account and administrator account. I had to reset the ownership and access so "everyone" could read and write. It is worth noting that re-installing BOINC does not remove and re-install the data directory, you have to do that yourself, and depending on the rights of the creating user you might end up with only the administrator being able to do anything with the folder (I think that's what I did wrong - but not with BOINC) One of the things I tried before posting was to uninstall BOINC and copy the BOINC data folder to a FAT32 formatted second partition on the same laptop, then I blew away the BOINC data folder and re-installed. Lastly, I copied the files back so this should have done two things: 1) FAT32 doesn't support NTFS Access Control Lists (ACLs) or Access Control Entries (ACEs), so those would have been removed from the files when copied to the FAT32 partition. 2) The thing to remember about NTFS permissions is "move on the same; retain." What this means is that NTFS permissions are only retained if the files are moved on the same NTFS partition. Any other action, like a copy on the same or a copy from another partition, would not move the NTFS permissions with the files. The files would inherit the permissions of the parent folder that they are being copied into. So what I did should have taken care of any permissions issue. However, the fact that the explicit running as Administrator has proven successful, there must be something somewhere that won't let it run in the standard user context even though such a configuration works well on every other machine in my house, including two other Vista machines. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I didn't get round to the problem machine today - for a few moments, it seemed that Patch Wednesday had claimed one of my W7 machines, but I managed to recover it. I'll try again tomorrow. My machine was supplied with Windows XP, and has continued to run it ever since. I have enabled the POS embedded security updates. It looks like the problem started one or even two months ago, but the machine is so slow (and reliable) that I haven't checked it in all that time. It's one of the very earliest P4s, and it's only remotely worth running for sentimental reasons - it was my mother's. And a credit to the build quality of early Dells. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
This is for you: |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Thanks. Looking through the logs, it appears that the problems started happening around 27 November last year. Oops. First problem was a task which ran on, and on, and on, until it eventually was cancelled for exceeding the time limit. The following tasks were aborted for 'not started in time'. After that (looks like there was a machine restart - was that me, or a power glitch?), tasks started crashing with 'output file absent' in the logs. Since the first thing that the SETI app does is to copy the WU header into the output file, that's a very early crash indeed. The rest you can read from the machine record I linked yesterday: <message> - exit code -1 (0xffffffff) </message> <stderr_txt> SETI@home error -1 Can't create file -- disk full? in seti_init_state() File: ..\seti.cpp Line: 394The disk is not full, but might have other problems. I'm going to do a controlled reboot with disk check, and take it from there. BOINC data folder is in All Users/appdata - where it's always been - so there should be no permissions problem. There have been periodic automatic security updates, but none between 11 October and 25 December. Most recent was 16 March, so more may be due. I'll check. More after lunch, perhaps. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Well, I'm not getting very far with this. The machine seems to be working OK (though even slower than I remember it): the hard disk checks out, no errors on a full pre-boot check. BOINC loads and runs OK - it's running as a service. SETI is using current software (SSE2 r3330). But the first task downloaded ran for several minutes with nearly zero CPU usage, then crashed as in my last post. The next three crashed the same way, but much quicker. I'm trying to return BOINC to user mode, but getting a consistent installer error 2203: The process cannot access the file because another process has locked a portion of the file. Eh? Never come across that one before. I've had a google round all the usual places, but nothing recognisable turns up. I'll sleep on it, but unless some bright spark turns up with a new idea, I'll probably give up on this machine - the speed of processing (best part of a day per WU) isn't worth the electricity. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
Whenever I come across a locked file that I can't figure out where the lock is, I use Unlocker , which can usually force unlocks on a file. Error 2203 sounds like a Windows Installer error and is most frequently caused by file/folder ownership or permission issues on %temp%. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
If it's XP attempting to run into safe boot could help. Apparently some process keeps lock on some of files BOINC needs during setup. Process Explorer/Disk Monitor could provide another hints. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I did try disabling the service-mode automatic BOINC run at startup, and repeating the installation attempt immediately after reboot, but still no success. Safe mode seemed harder than I remember to access - it's possible that some of the embedded-mode POS security updates have disabled some of the usual desktop facilities. I could try removing them, backwards, one by one - but I'm not sure the machine is worth the time. |
Kissagogo27 Send message Joined: 6 Nov 99 Posts: 716 Credit: 8,032,827 RAC: 62 |
F8 after bios POST and Before windows logo .. or MSCONFIG , general TAB and safe mode |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
All my F8 keys are well worn out - been there, done that. But this particular machine is behaving differently. |
bluestar Send message Joined: 5 Sep 12 Posts: 7031 Credit: 2,084,789 RAC: 3 |
The reason why you are not allowed to write to a BOINC lockfile is not because of privileges or administrative rights, but rather that a process should not be allowed to write to the file for another process, meaning adjacent. If there are two slots for two different processes, meaning tasks, if one was allowed to write to another for such a task, next you would always know the outcome, and therefore also the reason for it being so. Like a parent, also a sibling as well, and it is fun to notice that Windows has now grown into an environment where processes are the main way of keeping a tab with that of task processing itself. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.