05-17-2012 03:54 AM
I have a customer who have upgraded from EV9.0.2 to EV10.0.1 using the Migration Wizard and all went fine and is working as it should.
They wanted to convert indexes from 32 bit to 64 bit so as a test just tried it for a couple of users
It fails almost immediately every time for a convert or rebuild
Firewall and UAC turned off
Have followed advice on DCOM config rights, poisonpill and site level re-try errors.
It gives event 41352 (Index Volumes Processor)
The details in the event are:
Reason:Undefined
Event Type: Non Critical
Description: Retrieving the COM class factory for remote component with CLSID {3A92686F-E5E8-4505-ABB5-49E5F725617A} from machine evserver.ourdomain.com due to the following error: 800700005
A registry check shows this to be the Storage Crawler Service - have checked DCOM setting and set as per previsous advice (Everyone and anonymous have full control)
The task log shows it setting current live index to read only etc., estimating how manyitems to index, creating new hidden index, allocating the new index (I can see that the folder exists on the file system but has just a state.dat file in it of 1kb - nothing useful in it as far as i can see)
It then says adding items to the new index and immediately pauses due to an error - retries, pasuses, fails and gives the same details as in the event logs.
Any ideas please - thanks
Solved! Go to Solution.
05-17-2012 07:13 AM
Ok - this has been cracked and may help others in the future with similar issues when using the EV Migrator to move from EV9 to EV10 on a new server.
We found a lot of DCOM 10016 errors in the System Event Logs (one or more per second) relating to the NT AUTHORITY\ANONYMOUS LOGON
Although EV seemed pretty happy it was clearly having issues in the background with COM / DCOM issues.
This all came down to the SPNs were still set to the original server name that EV was migrated from.
So to resolve it, from a DC as a domain admin, run:
setspn /l oldservername
This shows that is has the ev server and evsite alias against it.
To remove it use:
setspn -D HOST/evalias oldservername
repeat for each alias
Next run:
setspn -A HOST/evalias newservername
repeat for each alais (ev server and ev site - short and fqdn depending on the results of you /l above)
Finally reboot the ev server for it to pick up the correct SPN setting
No more DCOM errors and Index conversions working - happy all around!
Hope this helps others
05-17-2012 04:46 AM
Have you see this technote? http://www.symantec.com/docs/TECH185042
05-17-2012 04:50 AM
Saw that one but no similarity to the event description or logs that we have
05-17-2012 07:13 AM
Ok - this has been cracked and may help others in the future with similar issues when using the EV Migrator to move from EV9 to EV10 on a new server.
We found a lot of DCOM 10016 errors in the System Event Logs (one or more per second) relating to the NT AUTHORITY\ANONYMOUS LOGON
Although EV seemed pretty happy it was clearly having issues in the background with COM / DCOM issues.
This all came down to the SPNs were still set to the original server name that EV was migrated from.
So to resolve it, from a DC as a domain admin, run:
setspn /l oldservername
This shows that is has the ev server and evsite alias against it.
To remove it use:
setspn -D HOST/evalias oldservername
repeat for each alias
Next run:
setspn -A HOST/evalias newservername
repeat for each alais (ev server and ev site - short and fqdn depending on the results of you /l above)
Finally reboot the ev server for it to pick up the correct SPN setting
No more DCOM errors and Index conversions working - happy all around!
Hope this helps others
05-17-2012 08:12 AM
good catch