New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
poor write performance with high measurement cardinality #13536
Comments
@kumud-pe it happens when you write large amount of new measurements/fields in different requests in a very short period. does it sound familiar with your case? |
Yes. We were trying to push Kafka metrics through a StatsD server. There will be a bunch of new measurements. But, it's not a delay in our case, the new measurements are not being written at all. Now, even if I try to write one new dummy measurement (manually), it's not getting written. |
i saw a 204 successful response in your log. i'm afraid it's not my case. in my case, the write request will hung for a long time until http timeout or an error. |
Oh. Ok. This could be a different issue then. Did you check disk I/O stats? That could be becoming a bottleneck. |
yep, i will submit a PR later for this |
i'm using 1.6.4, but i believe 1.7.x has the same behavior |
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. |
This issue has been automatically closed because it has not had recent activity. Please reopen if this issue is still important to you. Thank you for your contributions. |
when writing to a newly created database, with a high measurement cardinality, e.g. 100k, data set, the performance is poor.
after some digging, i found that the
MeasurementFieldSet
would synchronously overwrite and fsync the 'fields.idx' file within a mutex with each measurement/field update, which lead to all writing requests to queue up.one solution is to involve some delay sync techniq like the one used in WAL
The text was updated successfully, but these errors were encountered: