List:General Discussion« Previous MessageNext Message »
From:Vivek Khera Date:May 14 1999 1:25pm
Subject:Re: Killed processes / Server hanging
View as plain text  
>>>>> "MG" == Michael Griffith <mgriffith@stripped> writes:

MG> A week ago I posted a message about mysqld mysteriously taking 90% of the
MG> CPU time (and usually crashing) at random intervals. I know I've seen
MG> similar posts with no real solutions. I think I've found a clue and also at
MG> least part of the cause on my BSDI setup.

MG> Just to see what happened I ran 'unlimit' just before starting mysqld. It
MG> still crashes, but only half as often! Some process limit is causing mysql
MG> to hang up, but I can't narrow down to a specific resource or a specific
MG> query. The crashes now seem to be unrelated to CPU load.

Hmm. MySQL has been rock solid stable on BSD/OS 3.1 and 4.0 for me
ever since MySQL 3.21.13 or so, and I pay TcX $1000 per year to keep
it stable...  I run it under the user "mysql" which is in the login
class "daemon" so it gets sufficient resources.

When you type "unlimit" before you run mysql, what are the resource
limits you get?  They still may not be enough in terms of memory,
which could cause thrashing when a large (in terms of result set)
query is running.  You may need to increase the system-imposed hard
limits so you can get more resources (using the sysctl command.)

Thread
Killed processes / Server hangingMichael Griffith14 May
  • Re: Killed processes / Server hangingVivek Khera14 May
  • Killed processes / Server hangingMichael Widenius14 May