8. Checking for System Changes and Log Hints
The last step in this guide is considering any system changes and/or log hints.
Was there any operating system and/or application changes?
Check the Studio EVENTS page and
memsql.log
for any information about recent log events at a high level. Sort byevent_level
to see those that are critical.For more verbose logs, obtain the tracelog from the master aggregator data directory (the default for this is
/var/lib/memsql/node-nodeHash/data/tracelogs
).Find the
memsql.log file
and runtail -f
orcat
andgrep
for “error” and/or “warning”.
Final Recommendations
If nothing unusual arose in the investigation across all of this guide, but your cluster is still having performance issues, consider the following recommendations:
Did you properly capture the workload of interest in following the steps of this guide? It is possible an expensive query, queuing, or a resource problem was missed because the data wasn’t examined at the right time. Make sure that the workload of interest is running during the time of investigating.
Does your system have a recommended configuration? Check the
sdb-report
output from Step 1 again and consider any configuration optimization suggestions in the check output.Is the system’s hardware healthy overall? Check disk and network performance. It is possible that a hardware issue might be leading to performance bottlenecks.
Check out the Overview to see if any of the reference topics included there are applicable to your cluster.