0 Replies Latest reply on Jul 12, 2011 9:03 AM by twonk

    ePO 4.5.4 slowness/lock - hf1?

      We've been having trouble getting HF1 to install, so we haven't been able to test whether or not this resolves the problem. I'm hoping the community will read this and be able to offer insight from their experiences so we can prioritize our troubleshooting process better.


      We have a small number of EEPC clients (less than 200), so I would not expect this to be acing out the server. But, anything's possible, right?


      We're getting "Server is too busy (245 connections) to process request" in the server_error log. It happens frequently, but randomly, and is not associated with any specific task or product callback we can identify.


      This corresponds to errors in the orion log:


      ERROR [http-xxxx-Processor493] dataChannel.EPODataChannelMessagingService - Error writing DC packet to Work Queue

      java.sql.SQLException: Transaction (Process ID 1055) was deadlocked on lock | communication buffer resources with another process and has been chosen as the deadlock victim. Rerun the transaction.


      We're having trouble getting a database trace to correspond with the event as it happens - it looks like a different process every time - and I noticed the notes in HF1 list these as resolved issues:


      Issue: The use of products that depend upon the ePolicy Orchestrator Data Channel in large enterprises could cause the Agent Handlers and application server to become overwhelmed.

      The Data Channel functionality has been optimized to support large enterprises.


      Issue: If an ePolicy Orchestrator server was upgraded to 4.5 Patch 4 while there were pending Data Channel messages awaiting delivery, messages were no longer deleted from the database after they were delivered or expired. This caused a backup of messages queued for deletion and caused poor Agent Handler and ePolicy Orchestrator console performance.

      Resolution: The presence of pending Data Channel messages awaiting delivery no longer causes an issue when the ePO server is upgraded to a higher patch level.


      Has anyone else seen this, and if so, was HF1 the solution for you, or something else?