hansedong commented on issue #7353:
URL: https://github.com/apache/apisix/issues/7353#issuecomment-1175831285
@tzssangglass
I did a further follow up on this issue. The root of this problem is on Etcd.
When APISIX nodes establish more connections (such as more than 200) to the
same Etcd node, and APISIX communicates with Etcd through TLS certificates,
this problem can be reproduced.
This problem leads to 2 points:
1. When APISIX initiates an HTTP request to Etcd, Etcd's response time will
increase significantly (up to tens of seconds).
2. Because the APISIX Prometheus plugin will query the Modified Index of
Etcd in real time every time, the response time of the Prometheus plugin will
also increase significantly. Therefore, Prometheus will time out when fetching
metrics.
I have created a related issue in the Etcd project and reproduced the
problem. Related issues: #7078
https://github.com/etcd-io/etcd/issues/14185
Etcd version 3.5.5 will fix this issue. And, I rebuilt and deployed Etcd
based on its fixed PR and tested, and it has been confirmed that this problem
can be fixed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
