Video Screencast Help

Mail Archiving task Stopping/Restarting every 30 seconds

Created: 23 Aug 2013 • Updated: 31 Jan 2014 | 17 comments
The Boulon's picture
This issue has been solved. See solution.

Hello,

 

After having installed my cluster on a W2K8R2 I added 2 Exchange 2010 Mbx Servers. Installed EV 10.0.4 on the 2 nodes.

The task were created succesfully. But, when I check the event viewer, I can see that they are stopping then restarting every 30 seconds.

The problem is that the users archive manually. One time it's OK (when the task is up) and one time not (task is down).

 

Any idea to resolve that problem?

 

Thx

 

Olivier

Operating Systems:

Comments 17 CommentsJump to latest comment

Ben Watts's picture

When the Archive Task actually stops we should be throwing an error in the Event logs, do you see anything under the Symantec Event logs?

Also I would get a Dtrace running on the Active node for the ArchiveTask process to capture whatever is happening.  -  http://www.symantec.com/docs/TECH38122

 

Once you have that log we should be able to tell what is going on.

Pradeep_Papnai's picture

I had such issue privously where registry “RestartAllMAPITaskIntervalMins” was set with two hours (180 min) Please check the same registry at following location "HKEY_LOCAL_MACHINE\Software\KVS\Enterprise Vault\Agents\”
If you have that registry then either delete or increate the time upto 8-12 hours based on Support suggestion.

AS Ben told, need to take ArchivingTask dtrace to capture what exactly triggering restart of task.

Ben Watts's picture

Just looked through that, cant find anything abnormal, the shutdown (In EVs eyes) is expected, we throw a 3296 event to the logs, but again this is nothing abnormal for a task shutdown.

There are other things that the Tasks rely on, for example MSMQ, so if there is a problem with them then the tasks will shut down.

To be honest Im not 100% sure the RestartMAPI regkey will come into play here as it does what it says and literally restarts MAPI tasks not EV Tasks.
Something appears to give the Tasks a command to shutdown but we need to find what that is.

Is it all tasks that shutdown?
Do they all shutdown at the same time?
Do they all restart on their own rather than having a manual start carried out by an admin?
Have you recreated these tasks recently at all?
Do you see anything else in the Event logs at all?

Pradeep_Papnai's picture

I check the d-trace even I could not find out anything relevant which could cause task restart, just eradicating the possibilities of registry I told in my previous comment as that is not present in your system (verified registry info from the end of trace).

Ben MAPI task means EV task (mailbox & journaling) literally & sometime incorrect registry settings cause weird behavior, check TN http://www.symantec.com/docs/TECH68433

That registry is not present so we can move ahead. Adding what Ben asked

Do you have any script or batch job which restart task controller service or EV task?

Do you see any error message in application (such as event 1000)? You need to just take dtrace on task controller & archiving task only, just to avoid noise & size of log.

The Boulon's picture

I deleted the 2 original task created when adding the 2 exchanges and recreated one.

They did actually stop and start togheter at the same time.

Event viewer is clean. All blue.

 

I really do not understand. 

 

PS: I know that the customer has a firewall between vlan's. Bu the EV custer and Exchanges MBX are on the same vlan.

JesusWept3's picture

Any chance of getting a dtrace of TaskGuardian, TaskController and ArchiveTask altogether?

The Boulon's picture

I will do that monday when I am at the customer.

It's really frustrating because it's the last thing I have to solve.. all the rest is ok crying

The Boulon's picture

OK, here we go.

Attached, the trace Jesus asked.

 

Look at min 08:12

 

Also, strange is that GC jump.. or maybe I am wrong

 

 

AttachmentSize
dtrace-node2.zip 920.49 KB
The Boulon's picture

Probably sent the wrong trace.

Here is another one

AttachmentSize
dtrace_node2.zip 291.84 KB
JesusWept3's picture

Yup, you are absolutely right
It reloads the GC list, switches to a new one then tells all the tasks to shut down, restart to use the new GC

No idea why it would be doing this
You may want to consider opening a case with support and in the mean time use ExchangeGCOverride in the ExchangeServerEntry table just to get it stable

The Boulon's picture

Support called and expect a fix today :-)

Also forgotten to mention that they have a NetScaler here.

The Boulon's picture

I did try the GCOveride but the, I my tasks failes to start with a message that active directory cannot find my mbx server.

 

hmmmmmm

JesusWept3's picture

huh, so can you try GC://yourExch2010CASServer.domain.com/ instead?

Definitely an odd one, I don't really know how the netscalar works, but if its switching the contexts of the Global Catalog then I guess that makes sense, just not seen it work like that before

The Boulon's picture

i idid put GC://name_cas_array.domain.xxx

 

Support told me to reinstall Outlook and ESM ?!?

JesusWept3's picture

huh, really? so its pointing at an Exchange 2010 cas and saying it can't find the Exchange server?!?
how about the GC server mentioned in the other traces? VM126? they do seem to work, albeit for a few seconds! :)

SOLUTION
The Boulon's picture

Yep.. pretty strange.. but I believe it only impact the manual archiving.. importing a pst or scheduled archive works... strange...

 

I will ask a collegue to take a look at the netscaler and log it.