Java process takes 95% or above CPU time

[mrezzonico]

Hi all,

from time to time I am not able to connect with SWA (outlook-connector is ok).
After restarting scalix-tomcat all seems ok.
I realise that in my server the java process is above 100%.

According to the information found in
http://www.scalix.com/wiki/index.php?title=Scalix_11_FAQs#Java_process_takes_95.25_or_above_CPU_time
I have enabled the trace of the indexer but in the file /var/opt/scalix/NN/s/tmp/indexer.log I did not find important information.
Therefore I want to modify the parameters IDX_MAXLOAD and IDX_MINLOAD in the file /var/opt/scalix/NN/s/tmp/indexer.log
My server has 2 CPU's therefore I think that the following values are correct:
IDX_MAXLOAD=2
IDX_MINLOAD=0.5

Is this correct ?
Is there any other change that can solve my problem ?

Thanks in advance.
Miche

[mrezzonico]

I have modified these 2 values but the java process is still over 100%

Any help is appreciated.
Miche

[les]

I have modified these 2 values but the java process is still over 100%

Any help is appreciated.
Miche

Use the following....

IDX_MAXLOAD=6
IDX_MINLOAD=2

in /var/opt/scalix/NN/s/sys/general.cfg

These settings refer to load average per cpu, not the actual number of physical cpu's. So when load average goes beyond 6 it will stop the indexer and not resume until the load returns to below 2. I've never had issues with using these settings.
You will likely need to restart scalix and scalix-postgres after making the changes for them to take effect.

# Throttle the mailbox indexing service to avoid system crash
# where xx and yy are per-CPU load averages. The IDX_MAXLOAD setting
# will stop the indexer from indexing any further messages when the
# per-CPU load average goes above the defined value. The indexer will
# not resume indexing until the per-CPU load average goes below the
# value defined by IDX_MINLOAD.

[mrezzonico]

Hi les,

thanks a lot for your answer.
Yesterday, before reading you answer, I have deleted a mailbox e re-created it (there was a lots of errors in "fatal" referring to this mailbox).
I have run omscan in fix-mode and restarted the server too.
Now the "java" process seems ok.

Before making the changes in general.cfg I wait some hours and see if the problem is solved.

If you can help me again I have the following question:

Looking in scalix-sis-indexer.log, yesterday and today, I see a lots of messages like this (many lines each second):
2011-10-12 12:40:26,409 WARN [QueueManager] [BatchUpdater.performContentDeletes:270] No documents found to delete for DeleteContentInfo{IndexInfo{userID='05c000006d559eb4-911.2.1.291'}, indexID='760201c0-4be955d6-4e2d220e-120fd8a', parentDref='00011283df8539ce'}

Also in /var/log/apache/access_log I see a lots of messages like this (many lines each second):
192.1.2.118 - - [12/Oct/2011:12:40:33 +0200] "GET /sis/indexer?fn=delete&uid=05a000006d559eb4-911.2.1.291&pdref=000100489a224163&dref=0014eeefd158184f&indexid=760201c0-4be955d6-4c9c2ee5-3957b3 HTTP/1.1" 200 4 "-" "-"

I have read somewhere that the indexer can run for many days.
Looking in my old logfiles I see that in the past, only in May this year, there was a similar activity on my server.
Is this "normal" ?
Is there anything I can do ?

Looking in the directory /var/opt/scalix/l1/s/indexwork I see (now) 14194 files.
After 1min a less number of files ... and so on.
It seems me that these files are processed by the indexer.
Is it correct that when all these files are processed, the activity of the indexer will be "normal" ?

Thanks again.
Miche

[les]

Hi les,

thanks a lot for your answer.
Yesterday, before reading you answer, I have deleted a mailbox e re-created it (there was a lots of errors in "fatal" referring to this mailbox).
I have run omscan in fix-mode and restarted the server too.
Now the "java" process seems ok.

Before making the changes in general.cfg I wait some hours and see if the problem is solved.

If you can help me again I have the following question:

Looking in scalix-sis-indexer.log, yesterday and today, I see a lots of messages like this (many lines each second):
2011-10-12 12:40:26,409 WARN [QueueManager] [BatchUpdater.performContentDeletes:270] No documents found to delete for DeleteContentInfo{IndexInfo{userID='05c000006d559eb4-911.2.1.291'}, indexID='760201c0-4be955d6-4e2d220e-120fd8a', parentDref='00011283df8539ce'}

Also in /var/log/apache/access_log I see a lots of messages like this (many lines each second):
192.1.2.118 - - [12/Oct/2011:12:40:33 +0200] "GET /sis/indexer?fn=delete&uid=05a000006d559eb4-911.2.1.291&pdref=000100489a224163&dref=0014eeefd158184f&indexid=760201c0-4be955d6-4c9c2ee5-3957b3 HTTP/1.1" 200 4 "-" "-"

I have read somewhere that the indexer can run for many days.
Looking in my old logfiles I see that in the past, only in May this year, there was a similar activity on my server.
Is this "normal" ?
Is there anything I can do ?

Looking in the directory /var/opt/scalix/l1/s/indexwork I see (now) 14194 files.
After 1min a less number of files ... and so on.
It seems me that these files are processed by the indexer.
Is it correct that when all these files are processed, the activity of the indexer will be "normal" ?

Thanks again.
Miche

I've never looked that deeply into the indexer logs. Never had any issues. I would imagine that yes, once all files are processed that the indexer activity will be "normal". Of course though as new files are added the indexer does those etc. In reference to errors in the logs, i'd say most likely is that the indexer has a list of email to index, and before it can get around to indexing it, the user deletes the message and therefore an error gets logged as a reference is invalid.

I really don't think you've got anything to worry about.

[mrezzonico]

Hi les,

thanks !!
Looking in the directory /var/opt/scalix/l1/s/indexwork it seems me that in 11 hrs all files will be processed.

Miche