#5954 HIGH 8.2.0 (: oom-killer during olpc-update

[Zarro Boogs per Child]

#5954: oom-killer during olpc-update
-------------------------------+--------------------------------------------
   Reporter:  DanKrejsa        |       Owner:  cscott                              
       Type:  defect           |      Status:  new                                 
   Priority:  high             |   Milestone:  8.2.0 (was Update.2)                
  Component:  upgrade utility  |     Version:                                      
 Resolution:                   |    Keywords:  oom-killer olpc-update blocks?:8.2.0
Next_action:  never set        |    Verified:  1                                   
  Blockedby:                   |    Blocking:                                      
-------------------------------+--------------------------------------------
Changes (by ridderman):

 * cc: ridderman (added)
  * keywords:  oom-killer olpc-update => oom-killer olpc-update
               blocks?:8.2.0
  * priority:  normal => high


Comment:

 I've been experiencing this on multiple occasions for months, and just
 recently found this bug. About 50% of the time that I try olpc-update,
 Sugar restarts. I used to get MemoryError tracebacks on the console, but
 more recently it fails with a Sugar restart due to the oom condition. I'm
 guessing that this is not happeninng to many people, or else it would have
 been noticed and fixed by now. Here is my setup:

 G1G1, upgraded to several Update1 candidates, then many joyride images.
 When I run olpc-update, I su -l in the Terminal. I make sure that all
 activities are closed besides Terminal. I used to not know any better and
 would have one or two activities open. I think, anecdotaly, that if I've
 opened and closed other activites prior to the Terminal, the oom is more
 likely to occur. This might be because more memory is reserved for the
 system cache. Also, irsync_pristine fails every time on the file
 "localtime". irsync_dirty usually either works or the oom happens here. A
 few times, it's failed with a traceback and rsync has worked. I'll attach
 my /var/log/messages and sugar logs.

 If this is happening to all G1G1, then I would consider it an 8.2.0
 blocker. I've added the blocks?:8.2.0 keyword to get others' take on the
 situation.

-- 
Ticket URL: <http://dev.laptop.org/ticket/5954#comment:13>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system