SWA performance

[savant]

I had a similar problem with SWA being very slow after an upgrade from 10.x to 11.2.

After a bit of head-scratching, I noticed PostgreSQL wasn't starting on reboot. First I got the "can't bind to address" error, which I fixed by editing pg_hba.conf. When that was done, it still wouldn't start, because somehow the DB had become corrupted.

So, I used the script in /opt/scalix-postgres/bin, namely sxpsql-recreatedb to make a new one (after manually starting Postgre without any data). The script will fail if pgsql isn't running already, so make sure it is.

HTH,
Simon.

[chimney]

I should be more specific... Just reconfigure tomcat, and it'll ask how much ram to reserve. Just notch it up.

Kev.

I've added "CATALINA_OPTS="-Xmx128m"; export CATALINA_OPTS" to the Tomcat startup script and it didn't startup at all. Could you give me a hint on to reconfigure Tomcat to use more memory.

Chimney

[kanderson]

Rerun the scalix-installer script that you used to install Scalix initially.
Choose that you want to reconfigure the software (rather than install)
Choose to reconfigure (from the top of my head here) Tomcat.
When it asks how much RAM to assign, put in the appropriate number.

Kev.

[seany]

My only complaint about the web access is the startup time. Looking at the web source, it seems to be loading tonnes of seperate javascript and css files, each involving another server round trip.

Could the next version possibly tighted this up? I used to use GWT for making web content in a previous job and that showed exactly how to do it.

[a.schild]

The 11.3 release will have a faster loading web interface.

viewtopic.php?t=9208&start=15

André

[stevelogik]

Can you post your resolv.conf as well as the results of "hostname -f" and "cat /etc/hosts"?

Thanks
Kev.

My resolv.conf is the name server which is on another computer xx.xx.xx.x

hosts is

127.0.0.1 localhost.localdomain localhost
xx.xx.xx.x servername.domain.net servername

[stevelogik]

I have also re-ran the installer for tomcat and bumped up the memory. It is still slow. When I reboot the server it is really fast logging in, but 10 minutes later it is slow again. If I restart postgresql it is fast for 2 minutes and then slows down again

[stevelogik]

lastly when i do hostname -f i get
servername.domain.net

[kfleming]

seany's right, it's the heap of connections it's making to load all the javascript files. We made it slightly better by fiddling with Tomcat so it set the correct expires headers on all these files, so they have a better chance of being cached. I will track down the setting we changed and post it here.

[scalixuser]

[quote="savant"]
So, I used the script in /opt/scalix-postgres/bin, namely sxpsql-recreatedb to make a new one (after manually starting Postgre without any data). The script will fail if pgsql isn't running already, so make sure it is.

[/quote]
Thanks Savant!
This fixed it for me.
# service scalix-postgres start
# cd /opt/scalix-postgres/bin
# ./sxpsql-recreatedb

(Scalix 11.2.0/CentOS 4.4)

[STXRich]

Just wanted to ressurrect this thread to say that this fixed a long problem that we have been seeing with degrading SWA performance.

We were pulling our hair out. I kepts seeing multiple instances of the postmaster running in Top, and the server's %wa times were approaching 100%.

After running this, %wa is back to less than 3% and SWA is popping like it used to.

Scalix masters, is this something that should be added into the daily or weekly maint scripts? Is this a problem that you have come across before?
Is scalix-postgres not cleaning up after itself properly?

[florian]

in Scalix 11.4.3, we have finally implemented auto-vacuuming of the postgres db, so the database should behave much better now in terms of cleaning up after itself.

Florian.