Opened 14 years ago
Closed 14 years ago
#9829 closed defect (invalid)
'500 Internal Server Error' for timeline when logged in as a particular user
Reported by: | Owned by: | ||
---|---|---|---|
Priority: | normal | Milestone: | |
Component: | timeline | Version: | 0.12dev |
Severity: | normal | Keywords: | |
Cc: | david-sarah@… | Branch: | |
Release Notes: | |||
API Changes: | |||
Internal Changes: |
Description
http://tahoe-lafs.org/trac/tahoe-lafs is a Trac instance (running Trac 0.12.1dev-r9947) for the Tahoe-LAFS project. When logged in as a particular user account that has admin privileges, "davidsarah" (and apparently only then), the timeline page at http://tahoe-lafs.org/trac/tahoe-lafs/timeline always returns a '500 Internal Server Error'.
The bug ticket on the affected trac is http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1250 . (Comments 2..4 about hard drive issues are a red herring.)
I will post the relevant server log entries when I have them.
Attachments (0)
Change History (8)
comment:1 by , 14 years ago
comment:2 by , 14 years ago
Replying to david-sarah@…:
I will post the relevant server log entries when I have them.
The traceback from the log would be most helpful indeed.
comment:3 by , 14 years ago
It appears there is no Trac log entry generated by this error.
I made another request to an invalid URL before and after, which did cause a log entry, so presumably I'm looking at the right logfile (log/trac.log
in the Trac environment).
follow-up: 5 comment:4 by , 14 years ago
The weird thing is, the output from comment:1 is from Apache, not from Trac. When Trac generates a 500 internal error, it outputs a nice error page, or if that fails, a traceback as plain text.
Do you see anything related to the error in the Apache logs?
follow-up: 6 comment:5 by , 14 years ago
Summary: | '500 Internal Server Error' for timeline when logged in as a particular admin user → '500 Internal Server Error' for timeline when logged in as a particular user |
---|
Replying to rblank:
Do you see anything related to the error in the Apache logs?
The Apache error.log
says (wrapped for readability):
[Tue Nov 16 00:34:17 2010] [error] [client 90.217.140.169] FastCGI: comm with (dynamic) server "/usr/share/trac/cgi-bin/trac.fcgi" aborted: (first read) idle timeout (30 sec), referer: http://tahoe-lafs.org/trac/tahoe-lafs/timeline [Tue Nov 16 00:34:17 2010] [error] [client 90.217.140.169] FastCGI: incomplete headers (0 bytes) received from server "/usr/share/trac/cgi-bin/trac.fcgi", referer: http://tahoe-lafs.org/trac/tahoe-lafs/timeline
The Apache access.log
is probably not relevant, but anyway it says:
90.217.140.169 - - [16/Nov/2010:00:32:41 +0000] "GET /trac/tahoe-lafs/timeline HTTP/1.1" 500 810 "http://tahoe-lafs.org/trac/tahoe-lafs/timeline" "Mozilla/5.0 (Windows; U; Windows NT 6.0; en-GB; rv:1.9.2.12) Gecko/20101026 Firefox/3.6.12 (.NET CLR 3.5.30729; .NET4.0C)"
The fact that the "davidsarah" user has TRAC_ADMIN privilege is probably not relevant; removing that privilege has no effect on the error, and another user with TRAC_ADMIN does not have the problem.
comment:6 by , 14 years ago
Replying to david-sarah@…:
> aborted: (first read) idle timeout (30 sec),
I forgot to mention that the request takes about 38 seconds before failing, so that's consistent with a 30 second timeout.
comment:7 by , 14 years ago
So the FCGI server doesn't reply within 30 seconds, and Apache times out. It seems like Trac is choking on too much data for the timeline. Do you happen to remember setting a very high value for the "… days back" field in the timeline? This could explain the very long processing time, and the fact that it only happens to you and not other users.
Can you access the database directly? If yes, please post the output of:
SELECT * FROM session_attribute WHERE sid='davidsarah' AND name LIKE 'timeline.%';
A very high value for timeline.daysback
would explain the symptoms. You could set it to a small value with:
UPDATE session_attribute SET value='1' WHERE sid='davidsarah' AND name='timeline.daysback';
To avoid such an issue in the future, you should set [timeline] max_daysback
in trac.ini
to a reasonable value.
comment:8 by , 14 years ago
Resolution: | → invalid |
---|---|
Status: | new → closed |
Warner has confirmed on IRC that timeline.daysback
was set to 512. Reducing the value fixed the issue, so I'm closing this as an InstallationIssue.
The full HTML returned is: