[solved]issues with SWA after upgrade from 11.3 to 11.4

[alvaro]

Following an upgrade from 11.3 to 11.4 we have the following issues:

1.- the mobile web access page shows version 11.3 and users get an error when they try to log in: An error occurred: Unknown error (A00000)

2.- SWA page shows version 11.4, users can log in, but it is slower to display messages, specially when changing folders.

3.- Outlook connector is working fine, all problems we had with 11.3 are gone.

I have noticed we have two java folders in /usr/java: jre1.5_10 and jre1.5_13.


What would you do?

Thank you

[les]

Following an upgrade from 11.3 to 11.4 we have the following issues:

1.- the mobile web access page shows version 11.3 and users get an error when they try to log in: An error occurred: Unknown error (A00000)

Have you restarted httpd, scalix-tomcat etc?
Are all users having the problem or just some?

2.- SWA page shows version 11.4, users can log in, but it is slower to display messages, specially when changing folders.

Could there be heavy activity with postgres and sis indexing taking place?

I have noticed we have two java folders in /usr/java: jre1.5_10 and jre1.5_13.

That shouldn't matter. Only the upgraded path is looked at.

I have'nt seen any speed issues with the webclient post upgrade. Normally its much faster.

When you upgraded did you make sure that all users were logged out? there may be some defunct imap processes lying around.

it would be a good idea to log everyone out and restart the server (if you haven't already). Then we troubleshoot from a clean boot.

[alvaro]

Server has been freshly rebooted. Scalix installation has been run after reboot last night to repair Web Access and Mobile web access. Version on mobile web access now shows correct 11.4, however, users get same 'Unknown error' when logging in.. In 11.3 it all worked sweet.

Slow Web Access affects users with lots of emails in Inbox. 11.3 was quicker though.

Before running any upgrade we log users out and stop the Scalix Services.

How can I check postgres and sis activity?

Cheers
Alvaro

[/list]

[les]

Server has been freshly rebooted. Scalix installation has been run after reboot last night to repair Web Access and Mobile web access. Version on mobile web access now shows correct 11.4, however, users get same 'Unknown error' when logging in.. In 11.3 it all worked sweet.

Slow Web Access affects users with lots of emails in Inbox. 11.3 was quicker though.

Before running any upgrade we log users out and stop the Scalix Services.

How can I check postgres and sis activity?

Cheers
Alvaro

[/list]

look for sxmkindex or index.browse in a ps -ef or just by running top. If load average is low then thats probably not it.

It could be imap cache issues, try for one problematic user clearing the imap cache. There's a page on the scalix wiki about that. see if that helps.

[alvaro]

We have re-installed postgresql. S.
It appears something went wrong during the update and this errors were not logged.
SWA behaves very well. Fast authentication and fast changing folders.
Mobile SWA works now too, very fast.
Users happy.

[les]

We have re-installed postgresql. S.
It appears something went wrong during the update and this errors were not logged.
SWA behaves very well. Fast authentication and fast changing folders.
Mobile SWA works now too, very fast.
Users happy.

Well done! We like to hear that users are happy!!

[dyna]

I think i had the same problem. Also after the 11.3 to 11.4 upgrade.
I tried everything i could find in the forum/wiki, got rid off all errors, did some fine tuning and even upgraded some hardware but the problem kept returning.

After a reboot the whole system ran fine for a few days but then when someone tried to log into SWA it would freeze the entire systems for about a minute, even ssh login to the server was frozen.

I couldn't find any errors but did notice postgres and sometimes imapd had high cpu at these moments. I thought this was because the index rebuilding and played a lot with MAX/MINLOADS. The index processes also only grew and never disappeared over time as documentation says they should. Not even after 3 weeks with only a handful of users.

Was about to reinstall postgres when i noticed the new 11.4.1. Installed it and now everything seems fine. Now i also see the index processes actually doing something. Index seemed to be working before as indicated by indexer.log but it never really used any cpu let alone trigger the max/minloads. Now with 11.4.1 indexing does actually use the cpu and the max/minload values are triggered

SWA is now also much faster then it ever was in 11.4, even after a reboot with only 1 user logged in.

I Hope it keeps working this time. Thanks for the posting the solution.