Memory leak?

Continua CI Bugs
1

Hi!

Our Continua.Server.service.exe is running at over 6GB of memory. Granted, this is a beefy server, so it’s not really causing any problems, but it still seems a bit too much for just 17 days of up time and not a lot of activity. I’m attaching a memory log from WinDBG.__29137__0__continua_mem.zip (95.838 KB)

2

Hi Miruna,

Yes, 6GB is excessive. Unfortunately the WinDgb log does not give many clues as to the cause of this.

Which version of Continua are you running? Are any errors being reported in the event log. 

The beta version 1.5 has link to download a diagnostic report from the event log page. You can send this to support@finalbuilder.com. Otherwise can you provide some more information about your setup, e,g. number of projects, configurations, repositories, triggers, users, groups, notifications etc. 

3

Thanks, Dave! We were running 1.0.0.3016. We’re up to the beta now and the memory usage is at 2.3GB in after 2 days of running. I’ve sent the diagnostics log and the rest of the information to your Tech Support.

4

Hi Miruna,

It looks like the memory leak may be related to the LDAP synchronisation which is getting unresolvable SIDs errors when iterating through the members of certain built-in Windows groups. We have made some updates to the code to ensure that the objects are still disposed correctly in this situation and we plan to release a new beta build on Monday.

I also recommend that you use the groupsContainer attribute on the continua-configuration/authentication node in the service configuration file to restrict this search to groups within an LDAP organisational unit or container. 

eg.

<authentication mode=“Mixed” fqdn=“office.company.com” administratorsGroup=“Continua Administrators” groupsContainer=“OU=Continua,DC=Office,DC=Company,DC=com” />

This would prevent Continua from iterating unrelated groups. There are some timeouts in you event log which could be avoided by restricting the set of groups to iterate.

5

Thanks, Dave!

I’ve made the changes you suggested and we’ve been running for over a week now. The memory usage is at 450MB. I think we can call this issue closed.

Thanks again for your help.