-
Notifications
You must be signed in to change notification settings - Fork 1.9k
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
[BUG] opensearch crashes on closed client connection before search reply when total ops higher compared to expected #3962
Comments
@reta Just FYI, since you worked on the original issue. |
@dbonf Want to try to turn this into a unit test and artificially cause |
@dbonf thanks for details, I will take a look shortly
The
@dblock I added a test case as part of [1] which I thought is the same problem (turned out there is more than one flow). I will try to reproduce the exact failure (as per stack trace) although it is not easy. [1] #3626 |
@reta for example with few indices with one replica, two shards each and several thousand documents, few nodes, this query (designed to be slow to trigger the bug):
killed before the response triggers the crash, this is especially problematic when a client has retries in its logic so will kill many nodes in few seconds. |
@dbonf I think I deciphered the flow, what is happening:
Working on the test and thinking about the fix (not obvious at the moment). |
Describe the bug
Issue described in #3557 is still there also with opensearch 1.3.4.
In some circumstances, when running a search operation, if the client TCP connection is closed before the search operation is completed, opensearch crashes with:
The problem was supposedly solved with #3626 but we are still affected by it. The fix addresses the problem when
totalOps == expectedTotalOps
but not whentotalOps > expectedTotalOps
, when the exception is still unhandled and lead to a crash.To Reproduce
Steps to reproduce the behaviour:
We can consistently reproduce this behaviour in several environments.
Expected behavior
No crashes.
Plugins
Screenshots
no applicable
Host/Environment (please complete the following information):
Additional context
more logs:
The text was updated successfully, but these errors were encountered: