Zimbra Processor Output

[UltraFlux]

I was wondering if Zimbra actually requires all these reasources...

amavisd x 11
zimbra 22280 0.0 9.0 53296 45256 ? Ss 21:05 0:00 amavisd (master)
zimbra 22304 0.0 9.0 54024 45344 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22305 0.0 9.0 54024 45356 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22306 0.0 9.0 54024 45344 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22307 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22308 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22309 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22310 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22311 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22316 0.0 9.0 54024 45352 ? S 21:05 0:00 amavisd (virgin child)
zimbra 22317 0.0 9.0 54024 45348 ? S 21:05 0:00 amavisd (virgin child)

logger/mysql x 9
zimbra 22662 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22663 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22664 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22665 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22666 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22668 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22669 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22670 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22671 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/

mysqld x 4
zimbra 22668 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22669 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22670 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/
zimbra 22671 0.0 2.5 32668 12768 pts/3 S 21:05 0:00 /opt/zimbra/logger/mysql/bin/mysqld --defaults-file=/opt/zimbra/conf/

cyrus x 5
zimbra 23377 0.0 0.2 6536 1228 ? Ss 21:05 0:00 /opt/zimbra/cyrus-sasl-2.1.21.ZIMBRA/sbin/saslauthd -r -a zimbra
zimbra 23380 0.0 0.2 6536 1228 ? S 21:05 0:00 /opt/zimbra/cyrus-sasl-2.1.21.ZIMBRA/sbin/saslauthd -r -a zimbra
zimbra 23381 0.0 0.2 6536 1228 ? S 21:05 0:00 /opt/zimbra/cyrus-sasl-2.1.21.ZIMBRA/sbin/saslauthd -r -a zimbra
zimbra 23382 0.0 0.2 6536 1228 ? S 21:05 0:00 /opt/zimbra/cyrus-sasl-2.1.21.ZIMBRA/sbin/saslauthd -r -a zimbra
zimbra 23383 0.0 0.2 6536 1228 ? S 21:05 0:00 /opt/zimbra/cyrus-sasl-2.1.21.ZIMBRA/sbin/saslauthd -r -a zimbra

httpd x 5
zimbra 23424 0.0 1.0 13436 5328 ? S 21:05 0:00 /opt/zimbra/httpd-2.0.54/bin/httpd -k start -f /opt/zimbra/conf/httpd
zimbra 23425 0.0 1.0 13436 5328 ? S 21:05 0:00 /opt/zimbra/httpd-2.0.54/bin/httpd -k start -f /opt/zimbra/conf/httpd
zimbra 23426 0.0 1.0 13436 5328 ? S 21:05 0:00 /opt/zimbra/httpd-2.0.54/bin/httpd -k start -f /opt/zimbra/conf/httpd
zimbra 23427 0.0 1.0 13436 5328 ? S 21:05 0:00 /opt/zimbra/httpd-2.0.54/bin/httpd -k start -f /opt/zimbra/conf/httpd
zimbra 23428 0.0 1.0 13436 5328 ? S 21:05 0:00 /opt/zimbra/httpd-2.0.54/bin/httpd -k start -f /opt/zimbra/conf/httpd


What would happen if you had one process instead of 5 - 10 for each service?

[marcmac]

It depends on how much mail you're processing, of course.

Cutting amavisd down won't really cut down much on the processor usage, until you get a bunch of mail in - then you're basically choosing if you want to process all the mail for spam/virus in serial (fewer processes) or parallell (more processes).

You can cut down the logger sql connections (or turn off the logger, if you want) to save room.

Cutting down on the other sql connections will probably have a negative impact on perf - I'd do that last.

Overall, it's up to you - is it just you? Or are you designing a system for 500k users with heavy traffic and lots of simultaneous access?

[UltraFlux]

No its just for myself. I guess you could say I'm going for a Zimbra Lite type of approach. Keeping in tack all of the functionality like, loggin, spell check.

You mentioned reducing some of the services? What does one do to set these options?

cut down the logger sql connections

How do I do this?

I'd like to try the server with roughly half of the current process running and see how it works. Perhaps we might just end up creating Zimbra LiTE ^^

[marcmac]

For amavis:

Edit /opt/zimbra/conf/amavisd.conf and reduce the max_servers (it's at 10 now) down. 5 is a good place to start.

For logger:

Either edit /opt/zimbra/conf/my.logger.conf and reduce max_connections, then zmloggerctl stop, zmloggerctl start
OR
zmprov ms -zimbraServicesEnabled logger
zmloggerctl stop
to stop it completely. You'll get an error when you log into the admin console after this, and get no status or statistics in the admin console, but those aren't fatal.

You could also reduce max_connections in /opt/zimbra/conf/my.cnf and zmmailboxctl stop; zmmailboxctl start but that's not recommended.

If your spelling is good, you can shut down the spell server, and remove it from zimbraServicesEnabled much as the logger is above. Or, change /opt/zimbra/conf/httpd.conf:

StartServers 5
MinSpareServers 5
MaxSpareServers 10
MaxClients 150
MaxRequestsPerChild 50


Lower the first 3 (NOT the last 2) to cut down the number of apache processes running.