Dead2 commented on issue #1775: Deadlock in shm statistics
URL: 
https://github.com/apache/incubator-pagespeed-mod/issues/1775#issuecomment-394678320
 
 
   We do not have a snapshot of the logs from this deadlock unfortunately.
   
   We did consider turning statistics off completely, but that would cause 
ImageMaxRewritesAtOnce to be ignored and potentially turn into a DOS attack 
vector.
   
   I do wish there was a setting for disabling all "fluff" statistics, only 
keeping the few things needed for communication between processes. 
Investigating this also made me wonder why ImageMaxRewritesAtOnce is global 
while NumRewriteThreads and NumExpensiveRewriteThreads are process-local (so 
global if using a threaded apache and per-connection if using prefork). So this 
might be something worth changing long-term too.
   
   But I am confused, the documentation says NumRewriteThreads and 
NumExpensiveRewriteThreads are per-process, so those should not be able to 
cause a deadlock across processes. And ImageMaxRewritesAtOnce is global, so it 
could cause this, but two of the processes above are trying to rewrite .css 
files and should thus not be affected. So, either I am missing something or the 
documentation is misleading, probably the former. :)
   
   We will try out your patch, but proving the problem is fixed can take a 
while, since it can take weeks to trigger the bug (sometimes twice in a few 
days, sometimes once in a month, so it seems very random).

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

Reply via email to