#10075 NORM Not Tri: long uptime with constant wakeups fills /var/log and crashes machine

[Zarro Boogs per Child]

#10075: long uptime with constant wakeups fills /var/log and crashes machine
--------------------------------+-------------------------------------------
           Reporter:  rsmith    |       Owner:  cjb          
               Type:  defect    |      Status:  new          
           Priority:  normal    |   Milestone:  Not Triaged  
          Component:  distro    |     Version:  not specified
         Resolution:            |    Keywords:               
        Next_action:  diagnose  |    Verified:  0            
Deployment_affected:            |   Blockedby:               
           Blocking:            |  
--------------------------------+-------------------------------------------

Comment(by rsmith):

 Replying to [comment:1 pgf]:

 > just to be clear -- the log was full of the unavoidable kernel messages,
 happening for every suspend and resume, correct?  it wasn't your logging
 script?  (powerd may do some minimal logging at s/r time, but i it should
 be only one line or two.)
 >

 Yes. My logging script did not appear to be the culprit.  I did't study
 the var log messages in detail but when I looked at it I didn't see
 anything that stood out.

 > one s/r cycle per minute for 48 hours is about 3000 cycles.  18M of logs
 means over 6Mb per s/r cycle.  does this seem right?  i don't have a log
 handy to do an actual count on.

 Hmmm... seems excessive doesn't it. The machines can be restarted today
 and we can watch the logs to see what data is happening.  Perhaps
 something else is happening that generates a large amount of errors.

 More info I have a machine at my house that is running build 111 (w/ ohmd)
 + the same logging script sitting there and its not crashed.

 On that machine /var/log/messages however 6.9M

 Uptime says 1 day.  Kernel timestamp in messages (15000 seconds) is
 useless because time stops while suspended.

 Found a real time stamp at the start of the messages file when kernel
 boots.  3/17/2010 17:04:09 'date' now says 3/18/2010 18:49:15  so 7M in 25
 hours which is 78k per second.  Certainly something is up.

 > btw, even if logrotate configuration is tuned correctly for our reduced
 /var/log space (doubtful), i think it only runs once a day by default.
 even if your test was somewhat artificial, it seems like actual usage
 could give the same result after, say, a week or two of uptime.  i think
 the logging config should be examined.
 >

-- 
Ticket URL: <http://dev.laptop.org/ticket/10075#comment:2>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system