November 27, 2019 at 2:19 pm
Hello
I seem to have a moving target with SPs called via a number of ETL jobs
The reason I say it's a moving target is that this happens, maybe every couple of weeks (random) and it seems to be a different query each time.
Yesterday, a job that ordinarily takes less than an hour took in excess of 6
I've found the culprit - a single query that been stuck on 100% CXCONSUMER waits
Eventually, it cleared
Problem is, I'm struggling to find the source of the issue
Seems to be no blocking and nothing much else going on
CPU utilisation is around 60%
The only thing that bombs is page life expectancy although that's still aver 500
Any ideas where to look on this one?
Thanks.
- Damian
November 27, 2019 at 3:15 pm
Showing up as CXCONSUMER in solar winds
Yet, I have 12 suspended update CXPACET instances in sys.dm_exec_requests / sys.processes / sys.dm_exec_sql_text
- Damian
November 27, 2019 at 3:25 pm
CXPACKET and CXCONSUMER waits mean that your queries are going parallel... but not necessarily that the parallelism is a problem.
How do your ETL jobs work - with stored procedures or with ad hoc SQL? Query your plan cache to see which queries are using the most resources. If the queries are parameterised, consider whether parameter sniffing could be the problem. If they're not, consider whether you've got a lot of unnecessary recompiles. Give the query optimizer the best chance of getting the best execution plan by regularly updating your statistics.
John
November 27, 2019 at 4:46 pm
Thanks John.
ETL works with SP's
I'll take a look at some of your suggestions
- Damian
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply