2

Since our former data model is not very correct, the Slow queries panel shows that there are some queries which are performing slowly.

Slow Queries Screenshot

As I am planing to redesign the data model, I want to clear out the old information displayed in this panel, so I can see only information about my new data model. However, I do not know where OpsCenter is reading this data from.

My idea is that if this information is stored in a table or file, I can truncate or delete them. Or am I totally wrong with that assumption and this could be done by a configuration file modification or something similar instead?

OpsCenter Version: 6.0.3 Cassandra Version: 2.1.15.1423 DataStax Enterprise Version: 4.8.10

Adrian Sanguineti
  • 2,455
  • 1
  • 27
  • 29
feng1122
  • 119
  • 1
  • 2
  • 8

1 Answers1

2

It follows dse_perf.node_slow_log. Each node will track new events in the log as they occur, and store their top X. When viewing it in UI it gets the top X from each node and merges them. To "reset" you can truncate the log and restart the datastax agents to clear its current top X. There is a feature to reset for you in future but in 6.0.3 its a little difficult.

Chris Lohfink
  • 16,150
  • 1
  • 29
  • 38
  • I first truncate `dse_derf.node_slow_log`, and then restart the agents using `service datastax-agent restart` [page](https://docs.datastax.com/en/opscenter/5.1/opsc/reference/opscStartStopAgent_t.html) and OpsCenter using `service opscenterd restart`, but the slow queries information are still there... – feng1122 Nov 21 '16 at 05:40
  • That really should do it. Maybe worth trying again incase you missed a node or something. You can also try setting `slow_query_past` to 1 in the address.yaml and restarting agent. Then it wont try reading any past data when starting up, just follow from then on. – Chris Lohfink Nov 21 '16 at 07:11
  • Thanks a lot. I did the same procedure on another cluster which the version of OpsCenter is 5.2.4, and the Slow Queries information are disappeared. I think there is inconsistency between DataStax Enterprise Version: 4.8.10 and OpsCenter 6.0.3, am I right ? – feng1122 Nov 22 '16 at 02:26
  • 4.8.10 and opsc 6.0.3 should work well together. I cant think of a reason the procedure wont work unless the agent didn't actually restart with the service restart. You could try to `kill -9` the pid and start it again (making double sure the `node_slow_log` table is empty) – Chris Lohfink Nov 22 '16 at 02:34
  • I am sure the agent has been restarted correctly. When I executed `service datastax-agent restart`, I refresh the opscenter UI immediately, and I got the information 'Agent issues detected in xxx Cluster.' at the top of browser. – feng1122 Nov 22 '16 at 10:06