HTTP internal server error
7 Aug 2010 11:04:45 UTC
Topic 195226
(moderation:
Hello crunchers!
I have a problem:
Quote:
25/07/2010 11:42:40 Starting BOINC client version 6.10.18 for windows_intelx86
25/07/2010 11:42:40 log flags: file_xfer, sched_ops, task
25/07/2010 11:42:40 Libraries: libcurl/7.19.4 OpenSSL/0.9.8l zlib/1.2.3
25/07/2010 11:42:40 Data directory: C:\Program Files\BOINC_Data
25/07/2010 11:42:40
25/07/2010 11:42:40 Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q9300 @ 2.50GHz [x86 Family 6 Model 23 Stepping 7]
25/07/2010 11:42:40 Processor: 3.00 MB cache
25/07/2010 11:42:40 Processor features: fpu tsc pae nx sse sse2 mmx
25/07/2010 11:42:40 OS: Microsoft Windows Server 2003 "R2": Enterprise Server x86 Edition, Service Pack 2, (05.02.3790.00)
25/07/2010 11:42:40 Memory: 4.00 GB physical, 5.83 GB virtual
25/07/2010 11:42:40 Disk: 39.06 GB total, 10.41 GB free
25/07/2010 11:42:40 Local time is UTC +4 hours
25/07/2010 11:42:41 NVIDIA GPU 0: GeForce 8500 GT (driver version 19038, CUDA version 2030, compute capability 1.1, 256MB, 36 GFLOPS peak)
25/07/2010 11:42:42 Not using a proxy
...
07/08/2010 14:33:50 Einstein@Home Sending scheduler request: To fetch work.
07/08/2010 14:33:50 Einstein@Home Reporting 5 completed tasks, requesting new tasks for GPU
07/08/2010 14:34:00 Einstein@Home Scheduler request failed: HTTP internal server error
It's a problem with my client or BOINC server?
HTTP internal server error
)
It's definitely a problem on the server side.
There has been a similar problem at the SETI site (Message 1021602), but that was caused by obsolete application versions listed in the app_info.xml file, so it's very unlikely to be the same reason here.
Does the error occur also if you set No New Tasks and force an Update from the Projects tab in BOINC manager?
If so, could you identify (one of) the tasks trying to report? Perhaps that would give more hints.
I just saw that the server logs are accessible again, so you could click the "Last contact" link for host 1275368 on the Computers belonging to hoarfrost page after the update, to see if you see messages correlated with the internal error.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
Uh yeah, but the problem is
)
Uh yeah, but the problem is that there won't be a report generated as he can't reach the server. The error itself means as much as this: His computer knocks on the door at the server, and instead of being pointed left by the CGI scheduler to where the information is, it's pointed right into a dead-end where there's nothing.
The error is generated by the client (by the libcurl module) as soon as the signal seems to get lost. Is one of the mirrors down, or having problems?
I had a contact to upload & report work at around the same time which went through without a hitch. I think I used the Glasgow mirror, but am not sure. (Is that info somewhere in one of the sched_*.xml files, as I cannot find it so quickly)
RE: Does the error occur
)
Yes, it'is works.
But requests of new tasks failed until I exclude receiving a GPU-tasks in my account. Now works fine:
Thank you! :)
For the sake of debugging
)
For the sake of debugging this problem could you re-enable GPU tasks, issue one work request (update project) and make the sched_request_einstein.phys.uwm.edu.xml available to me (post it here, put somewhere to download or send via PM)?
Also the messages "... already reported as success" indicate that the first part of the scheduler contact (reporting results) had been successful and the error happens in the scheduler itself. So there should be a partial scheduler log. Could you post the id of your host?
BM
BM
RE: Could you post the id
)
It should be this one.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
I'm having the same problem.
)
I'm having the same problem. My Task is not able to report home to say that it's done.
RE: For the sake of
)
It might be worth having a look at the code round about the area where changeset [trac]changeset:22090[/trac] was made: that's the fix which solved the similar SETI problem that Gundolf referred to.
RE: I'm having the same
)
You can select the Projects tab in your BOINC manager, click the "No new tasks" button and then the "Upload" button. That would at least let your client acknowledge the already reported tasks. You'll get messages that the tasks already had been reported, but that's no error.
Since your computers don't have CUDA devices, I don't know where the problem originates, but one of your computers (3201971) obviously has its preferences set to the effect that no disk space is available. See server log:[pre]Request: [USER#509970] [HOST#3201971] [IP 122.164.160.158] client 6.10.58
[send] No disk space available: disk_max_used_gb 0.00GB disk_max_used_pct 20.00 disk_min_free_gb 0.00GB
[send] No disk space available: host.d_total 40.01GB host.d_free 1.50GB host.d_boinc_used_total 0.22GB
[send] No disk space available: x1 0.28GB x2 7.78GB x3 -0.00GB x -0.00GB[/pre]
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: 07/08/2010
)
I dug out the scheduler log of this contact, it looks fine, i.e. the scheduler ran (0.13s) without crashing.
I don't yet know what happened.
BM
BM
RE: I'm having the same
)
Could you post the ID of your host, and possibly the timezone it's set to, so I could look up the corresponding scheduler and httpd logs?
BM
BM