Limit the number of dedupe jobs running in parallel (7-Mode)

storageadmin · ‎2016-05-12

In 7-Mode only schedules for dedupe jobs are possible beginning at full hours e. g. 8:00am 10:00pm etc. Having numerous jobs it leads to situations where up to 8 jobs are running during the first minutes of an hour whereas the last minutes of an hour are rarely used for dedupe jobs. Although running with low priority we found 8 jobs do have an impact on performance. So we found it useful to limit the number of running dedup jobs by setting the hidden option "sis.max_active_ops" to a value of e. g. 6. This leads to a more even distribution of the jobs.

Using this be aware of the following caveat: http://mysupport.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=833716

Up to now we experienced no problems with this - even without booting. Any experiences of others on this subject?

Jeff_Yao · ‎2016-05-17

PLEASE REBOOT THE SYSTEM ASAP... without reboot, the stale-dip can't be initialized to right size, which will cause memory leak, ultimately data corruption.

BTW dont offline any volume where sis op is running on.

thanks

Jeff

storageadmin · ‎2016-05-27

Hi Jeff,

thank you for your comment. Meanwhile the systems have been booted.

Regarding your second comment with offlining volumes: Is this recommendation related to the situation before boot or generally

to the situation if the option doesn't have its default value? -

Should offlining be avoided only during actively running sis jobs on a volume or generally on volumes with configured sis?

Thomas

Jeff_Yao · ‎2016-05-27

sorry for the confusion.

I mean "during the interim 'after you change that option' and 'before reboot', DO NOT offline any volume with sis running on it". after you reboot ur system, it should be fine now.

thanks and have a nice weekend ahead

Jeff