[Thinlinc-technical] Do vsmserver and vsmagent renice after a few days?

Michael Bruns bruns at mps.mpg.de
Thu Mar 19 13:30:39 CET 2015


On 03/19/15 11:18, Patrik Pira wrote:
> Den 2015-03-19 08:06, Peter Astrand skrev:
>> On Tue, 17 Mar 2015, David G Miller wrote:
>>
>>> Subject says it all.  We have noticed that vsmserver and vsmagent
>>> start with a nice of 0 (normal Linux priority) and then apparently
>>> renice to a nice of 10 after a few days.  This creates problems for
>>> users who are starting processes since the child processes that they
>>> start inherit the nice setting of the parent process.
>>
>> We haven't heard of this problem before, but after some investigation, I
>> found this bug report:
>>
>> https://bugs.launchpad.net/ubuntu/+source/apache2/+bug/19797
>>
>> Are you using Ubuntu? Which version? Unfortunately the Ubuntu folks
>> seems to think that this is not a high priority bug...
>
> Same problem on RHEL6,
>
> this is due to anacrontab running cron.daily with "nice". cron.daily
> runs logrotate which upon rotating logfiles does a condrestart on
> vsm{agent|server} while running with the default niceness of 10.
>
> Edit /etc/anacrontab and remove "nice" on cron.daily should fix it.

same on SLES11/12 but with nice 15:

/etc/crontab:
...
-*/15 * * * *   root  test -x /usr/lib/cron/run-crons && /usr/lib/cron/run-crons >/dev/null 2>&1

/usr/lib/cron/run-crons:
...
CRON_SCRIPT_NICE_VALUE=15
...
nice -n ${CRON_SCRIPT_NICE_VALUE} $SCRIPT >$TEMP_MSG 2>&1

Is it really necessary to completely restart vsm{agent|server} or is some kind of reload
imaginable?

Michael.


More information about the Thinlinc-technical mailing list