[Bug 5731] DB_File locking hangs on Ubuntu feisty / gutsy Linux systems in spamd

[bu...@bugzilla.spamassassin.org]

http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5731


jm@jmason.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|DB_File locking hangs on    |DB_File locking hangs on
                   |Ubuntu feisty / gutsy Linux |Ubuntu feisty / gutsy Linux
                   |systems                     |systems in spamd
   Target Milestone|Undefined                   |3.3.0




------- Additional Comments From jm@jmason.org  2008-01-18 01:35 -------
I added that timeout code to the Bayes tie() calls as well; it's
in the current rev, r613114, and has been for a few days.  However, even
with that, last night a buildbot test run failed with this bug again:


http://buildbot.spamassassin.org:8010/t-feisty-561/builds/749/step-test/0

[28227] dbg: rules: running uri tests; score so far=6.909
[28227] dbg: rules: compiled uri tests
[28227] dbg: rules: running rawbody tests; score so far=6.909
[28227] dbg: rules: compiled rawbody tests
[28227] dbg: rules: running full tests; score so far=6.909
[28227] dbg: rules: compiled full tests
[28227] dbg: rules: running meta tests; score so far=6.909
[28227] info: rules: meta test T_TVD_PH_BODY_META_ALL has dependency
'TVD_PH_BODY_ACCOUNTS_PRE' with a zero score
[28227] dbg: rules: meta test T_FAKE_REPLY_SURE_A has undefined dependency
'FAKE_REPLY_A'
[28227] dbg: rules: meta test T_FAKE_REPLY_SURE_A has undefined dependency
'FAKE_REPLY_A2'
[28227] dbg: rules: meta test T_FAKE_REPLY_SURE_B has undefined dependency
'FAKE_REPLY_A'
[28227] dbg: rules: meta test T_FAKE_REPLY_SURE_B has undefined dependency
'FAKE_REPLY_A2'
[28227] dbg: rules: meta test DIGEST_MULTIPLE has undefined dependency 'DCC_CHECK'
[28227] dbg: rules: compiled meta tests
[28227] dbg: check: running tests for priority: 1000
[28227] dbg: rules: running head tests; score so far=9.225
[28227] dbg: rules: compiled head tests
[28227] dbg: rules: running head_eval tests; score so far=9.225
[28227] dbg: locker: safe_lock: created
./log/user_state/auto-whitelist.lock.vmsa.28227
[28227] dbg: locker: safe_lock: trying to get lock on
./log/user_state/auto-whitelist with 0 retries
[28227] dbg: locker: safe_lock: link to ./log/user_state/auto-whitelist.lock:
link ok
[28227] dbg: auto-whitelist: tie-ing to DB file of type DB_File R/W in
./log/user_state/auto-whitelist
[28227] dbg: generic: starting tie timeout at Thu Jan 17 20:48:25 2008

Maybe you need to kill a running spamd process?
started at 1200631702, gave up at 1200631831

# Failed test 1 in t/spamc_optL.t at line 20
	Not found: learned spam = Message successfully un/learned

Now, the tie timeout is for 30 seconds.  The tie() started at Thu Jan 17
20:48:25 2008, and the test harness gave up at Fri Jan 18 20:50:31 2008.  
That's a diff of over 2 minutes. So it appears that that timeout code won't
work, since a simple SIGALRM doesn't interrupt the hang. :(


PS: interestingly I haven't seen a single instance of this that *wasn't*
in a spamd-based test.  updating summary.





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.