-
Notifications
You must be signed in to change notification settings - Fork 24.8k
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
Add Circuit breaker on Transport ResponseHandlers #66196
Comments
Pinging @elastic/es-distributed (Team:Distributed) |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
I don't think there's a general transport-level solution for limiting the size retained by a response handler since the memory retained by a handler is pretty much independent of the transport layer. The OP doesn't tell us anything about what requests correspond with the problematic response handlers, nor what version they're using. Likely culprits include searches (in which case this duplicates #67478), stats (in which case this duplicates #55550) or bulks (in which case this is resolved by the new indexing pressure mechanisms). @easyice to what requests do the problematic handlers relate? |
I can't agree any more @easyice is my coworker, The cluster that this issue mentioned is the same to me ES version: 6.8.0, The request were bulk in the last two accident @DaveCTurner |
Ok, there is a solution for limiting the size retained by bulk response handlers, and it's already implemented (as of 7.9) so I think there's no further action needed here. |
@DaveCTurner Thanks for reply, Let me add the relate requests, It doesn't just appear in bulk,also in search request,like this: |
Indeed, we're tracking the issue for search responses at #67478. |
for every in flight Transport request, we add a handler at
org.elasticsearch.transport.Transport.ResponseHandlers
some times, the users sent a lot of request, such as query, that they exceed the system's capacity,then, CPU utilization reached 100%, but ,The client still sends a large number of requests,the requests will be add in
ResponseHandlers#handlers
then the Elasticsearch nodes will be oom
This happened to me a few times,I dump the jvm memory,open with Eclipse Memory Analyzer, it show the
ResponseHandlers#handlers
used 7.7GB memoryThe text was updated successfully, but these errors were encountered: