|
If you were logged in you would be able to see more operations.
|
|
|
Hyperic HQ
Created: 03/May/07 09:24 AM
Updated: 09/May/07 12:09 PM
|
|
| Component/s: |
Server
|
| Affects Version/s: |
3.0.4
|
| Fix Version/s: |
3.1.0,
3.0.5
|
|
| Last comment: |
80 weeks, 2 days ago
|
| Resolution Date: |
04/May/07 08:41 AM
|
|
It has been observed in large environments where the lather settings are set higher than the default (25) that the server can get locked up processing backlogged data from agents. The amount of data that is backlogged does not seem to matter. I have seen this with the server only being down for about 2 minutes.
A full thread dump shows contention in the sending of metric events after the data has been persisted.
I'll attach the full thread dump.
|
|
Description
|
It has been observed in large environments where the lather settings are set higher than the default (25) that the server can get locked up processing backlogged data from agents. The amount of data that is backlogged does not seem to matter. I have seen this with the server only being down for about 2 minutes.
A full thread dump shows contention in the sending of metric events after the data has been persisted.
I'll attach the full thread dump.
|
Show » |
|
Bug
Closed
Blocker
High load and transaction timeouts on server restart when lather connections setting is high
See attached thread dump. There are 2 separate thread dumps each taken about 5 minutes apart. The server also had verbose gc logging turned on, so you can ignore those entries.
There are dozens of entries stuck on:
org.hyperic.hq.measurement.server.session.DataManagerEJBImpl.sendMetricEvents