Users can't enter site in Share for a short period of time?

[gronfelt]

I'm running Community 3.2 with multi-tenancy enabled and from time to time my users can't enter specific sites in Share. The problem first appeared yesterday and after a few tries during 5-10 minutes everything seemed to work. Now the situation is the same again, this morning our users could not enter some sites, after 5-10 minutes everything works as normal.

Sites created after the problem first appeared yesterday is no problem, only older sites.

The logs show nothing at all.

The problem occurred about the same time of day today and yesterday, is it possible that some sort of periodic maintenance procedure is causing this (either in Alfresco or in the hosting Linux system)?

[kevinr]

This looks like it might be a regression. It sounds *very similar* to a bug that occurred in one of the first versions of the Share application and was quickly fixed. I have a feeling (i am guessing) that it is related to the changes made to Share to support Multi-Tennant - which annoyingly also broke NTLM support for Share - so I suspect this is actually the same problem under the covers, and has caused this regression also. I will see if we can investigate ASAP.

Thanks,

Kevin

[kevinr]

Can anyone who has seen this please tell me what their setup is, i.e. Alfresco/Share versions, are they on separate Tomcats instances or same one, server OS type, authentication used for share/repo etc. as this will *really* help us in finding out if this -is- a regression or not. As i have found the old bug relating to this:
https://issues.alfresco.com/jira/browse/ETHREEOH-806
but it only occurred when using NTLM on Share so it may not be the same issue…

I've asked around and none of the engineers nor QA have actually seen this issue occur locally so as much information as you can provide would be great thanks!

Kevin

[jtp]

New install of 3.2 Community on Win2003.  One Tomcat instance.  Alfresco is SSO Passthru with LDAP Sync.  Share is manual authentication passthru.  Share would be SSO, but as you mentioned, there's a bug.  Gronfelt and I are the major posters in that thread, so I'm not surprised to see us having similar issues here.  Anything else I can provide?  If you need to see my authentication subsystems configuration let me know.

[paolo]

From jira:

Kevin,
we are running Alfresco Community edition 3.2 on a unique tomcat instance.
It's running on CentOS 5 with Alfresco Ntlm authentication.
Please let me know if you need more details on our system.

Paolo

[kevinr]

Paolo, are you using Multi-tennant?

[kevinr]

Looks like everyone who has seen this is using some form of Alfresco SSO Passthru - be it NTLM or LDAP…? So I'm starting to wonder if this regression has been caused by the refactoring of the sub-system authentication on the repository side (which is significant as the fix to the original bug required some notable changes to the repository client auth layer - which has changed a lot since the sub-system refactor…)

Thanks for any further info!

Kev

[paolo]

No we are not using multi-tenant.
Just consider that our system is pretty much similar to a default Alfresco 3.2 installation since we didn't change a lot since the first installation.

[paolo]

Kevin we are using the default alfrescoNtlm authentication that come with the first installation. No changes on the authentication side.

[gronfelt]

We're running an installation that was originally the standard 3.2 release, since then it has been rebuilt with sources from the HEAD. This problem occurred both before and after rebuilding.

We're using LDAP authentication only, against OpenLDAP with an authentication chain with two LDAP instances. We've also enabled Multi-tenancy.

Server is CentOS 5.3.

[kevinr]

Alfresco QA have confirmed the issue. I'll let you know when we have more details.

Kev