[DocDB] The statistic threads_running_thread_pool for YCQL shows incorrect value #13575
Assignees
Labels
area/docdb
YugabyteDB core features
kind/bug
This issue is a bug
priority/medium
Medium priority issue
Comments
fritshoogland-yugabyte
added
area/docdb
yugabyte-ci
added
kind/bug
|
@rthallamko3 , something to add to the tooling bucket. I'm also intrigued this shows up now, but didn't before, perhaps some components were moved around, leading to some reporting on the YCQL webserver? |
|
I noticed this for a very long time, but forgotten about it.
When I created yb_stats, I had a hard time dealing with it, and was glad I found a way to ignore this.
I guess this is showing these values as long as I know...
Frits Hoogland | Developer Advocate
Mobile: +31-6-14180860
https://www.yugabyte.com
… On 11 Aug 2022, at 16:17, Bogdan Matican ***@***.***> wrote:
@rthallamko3 <https://www.google.com/url?q=https://github.com/rthallamko3&source=gmail-imap&ust=1660832261000000&usg=AOvVaw2T8kprH7wObYmwCZKyYp26> , something to add to the tooling bucket. I'm also intrigued this shows up now, but didn't before, perhaps some components were moved around, leading to some reporting on the YCQL webserver?
—
Reply to this email directly, view it on GitHub <https://www.google.com/url?q=https://github.com/yugabyte/yugabyte-db/issues/13575%23issuecomment-1212050804&source=gmail-imap&ust=1660832261000000&usg=AOvVaw31tAWsr7F1IvC7lu-_jpTf>, or unsubscribe <https://www.google.com/url?q=https://github.com/notifications/unsubscribe-auth/ATP5QQWTODAS4S6BJPZHA4DVYUDQJANCNFSM56IISUGA&source=gmail-imap&ust=1660832261000000&usg=AOvVaw2zKRgJ0tlPmK9J7Ip7qHtq>.
You are receiving this because you authored the thread.
|
|
@yusong-yan , This is a good ramp up item about metrics. |
|
This seems to have gone silent? |
I already have the fix, and it is in the process of reviewing. Sorry for the delay! |
|
Sorry! Is there a way I could have found this out myself? |
|
@fritshoogland-yugabyte For work in progress, you'd have to check Phabricator, or be explicitly added on those diffs as a subscriber. See https://phabricator.dev.yugabyte.com/D19359 |
yusong-yan
added a commit
that referenced
this issue
Oct 20, 2022
Summary: To track threads_started_category and threads_running_category correctly, rewrite ThreadCategoryTracker. The previous implementation can only take one metric_entity from multiple servers. It means when a new metric entity comes in, the tracker will replace old metric entity with the new one. For example, after creating tablet_server, ThreadCategoryTracker will track threads for yb.tabletserver. When the cql_server metric gets created, ThreadCategoryTracker will stop tracking threads for yb.tabletserver, instead track it only for cql_server. This is also the root cause of the negative metric value for thread_running_category in cql.server as we see in GitHub issue [[ #13575 13575 ]]. The new version of ThreadCategoryTracker is able to register a unique copy of these metrics in multiple servers. Plus, it uses function gauges which have the same style as how we track other metrics in thread.cc, such as cpu_stime, threads_started, voluntary_context_switches. Test Plan: Tested on local machine, both 12000/prometheus-metrics and 9000/prometheus-metrics have the same metric values of thread_running_category and thread_started_category Reviewers: amitanand, rsami, rthallam Reviewed By: rsami, rthallam Subscribers: bogdan, ybase Differential Revision: https://phabricator.dev.yugabyte.com/D19359
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Jira Link: DB-3154
Description
When querying the metrics for the YCQL (12000) endpoint, I get the following values:
These do not seem like realistic numbers to me, and therefore give the impression this statistic is broken.
I now find this value with version 2.15.2.0-b83, but witnessed this value for quite some time.
The text was updated successfully, but these errors were encountered: