[ 
https://issues.apache.org/jira/browse/KUDU-3021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16995734#comment-16995734
 ] 

Alexey Serbin edited comment on KUDU-3021 at 12/13/19 11:28 PM:
----------------------------------------------------------------

What's definitely missing is the histogram metric to see how big incoming 
transactions are.  I think that's useful in either case. :)

I think it's also useful to have a separate metric to track just the 
transaction-memory-limit-related rejections (i.e. not counting in the limits of 
the ancestral memory trackers).


was (Author: aserbin):
What's missing is the histogram metric to see how big incoming transactions are.

> Add metric gauges for the size of transactions applied to tablets and number 
> of rejected transactions due to their size
> -----------------------------------------------------------------------------------------------------------------------
>
>                 Key: KUDU-3021
>                 URL: https://issues.apache.org/jira/browse/KUDU-3021
>             Project: Kudu
>          Issue Type: Improvement
>          Components: master, tserver
>            Reporter: Alexey Serbin
>            Priority: Major
>              Labels: observability, scalability, troubleshooting
>
> Kudu servers have a limit on the size of transaction applied to a tablet: 
> {{\-\-tablet_transaction_memory}}
> It would be nice to introduce corresponding metrics to gauge how relevant the 
> current setting is for the actual size of transactions applied in a running 
> Kudu cluster. That can help with pro-active tuning of a Kudu cluster to 
> sustain a planned increase in workload.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to