Closed joshuaspence closed 7 years ago
Something definitely seems to be wrong with Logstash 2.4.0. I have rolled back our Logstash hosts to version 2.3.4. You can see the impact that this had by looking at the rate of messages published into SQS:
(Ignore the last data point, which always seems to be zero)
@ph could just be related to recent changes in the sqs plugin?
The output from bin/logstash-plugin list --verbose
shows the same version of logstash-output-sqs
on version 2.3.4 and 2.4.0.
yeah, one that changed is beats, which was also PH's work, so I'll keep him on the loop
@jsvd I doubt the SQS is in cause it havent been updated/changed for a few months, the obvious choice would be the beats inputs since we did a rewrite, in our test we had not experienced any memory leak but we may have miss something.
I think we have 2 problems:
Once I get the heap dump from #5932 I could see if its related or not.
@joshuaspence the data spike is something we need to look into.
A few question that come to mind:
What do you mean by duplicates? I was watching the logs on the Filebeat client and didn't see anything out of the ordinary except that a few times (not too often) I saw an error along the lines of "Connection reset by peer".
@joshuaspence I was just wondering if you had a lot of errors, transmission errors in FB will often generate duplicates.
Not that I could identity, no.
@joshuaspence Would you mind updating to the lastet input beats? I've fixed an issue concerning the shutdown that we have encountered when running the integration test and it look similar to your problem. At the time of writing the latest version is 3.0.4
v3.0.4 seems to be from July. Did you mean v3.1.6?
@joshuaspence oops yeah I meant 3.1.6, current 3.1.12 is this still an issue for your?
Nope, we are no longer seeing this issue.
We have just recently updated Logstash from version 2.3.4 to 2.4.0. It seems that since updating Logstash we have been having throughput issues (I suspect we are dropping logs somewhere, either in Filebeats or in our Logstash shippers). I noticed that running
service logstash restart
has been consistently failing after the upgrade. Additionally, when I attempt to restart the service I see the following errors appear in the logs:This issue might be similar or related to #5932. Our Logstash configuration is: