ArchiveOne Enterprise, all versions
A known issue has been found with ArchiveOne whereby in some customers environments the service can run out of NSPI connections. This causes the following symptoms:
- Policies may fail every few days with 'unable to connect to Exchange', until the service is restarted.
- Retrievals may begin failing for all users until the service is restarted.
- The Administration Console may fail to connection to the ArchiveOne server, until the service is restarted.
The root cause of this issue is a change in the operation of Microsofts RPC layer code that is used when the ArchoiveOne service is connection to a Domain Controller to gather information about mailboxes.
The change is causing resources to be used up and not released correctly, and eventually those resources reach the limit, no more can be created and the RPC connections connections fail.
This issue is actively being pursued by Barracuda, working closely with Microsoft to find a resolution.
The advice in the meantime, to reduce the frequency of this problem, is as follows:
- Raise the default number of allowed NSPI connections as detailed in this article: http://support.microsoft.com/en-us/kb/949469 (Barracuda recommend raising this to 500, however experimentation is advised for customers to find a suitable value that works in their environment and does not affect other infrastructure/applications).
- Restart the ArchiveOne service at regular intervals (Barracuda recommend every 2 to 3 days, however customers should find a suitable schedule that works for them, taking into account the policies they run and the time/schedule of those).
NOTE: a simple scheduled task executing "net stop AOnePolService && net start AOnePolService" should suffice.
The Barracuda development TFS Field Fix ID is 15706 and the Microsoft Support Case ID is 115011412266703
When this issue is resolved, a new hotfix build will be produced (if Barracuda changes are required) and made available to all customers.
Link To This Page: