João Duarte
João Duarte
Any deployment out there consuming from Agent or Beats with a heap usage above 75% (or even 50%) can potentially crash on upgrade as any direct memory allocation from beats/agent/http/tcp...
The netty errors come from the elastic_agent input, so we can disregard the rest of the stack for this issue and concentrate on Agent connecting to Logstash. `Received fatal alert:...
replaced by #16592
Opened https://github.com/elastic/logstash/issues/17147 to track the missing files on the Logstash image.
I've beefed up the alternative design section with other options: non-mfa api keys, totp, and a broader "allow-all" alternative, given ruby gems don't have namespaces.
The heap dump shows a single instance of org.jruby.ext.thread.Queue has 383.060 nodes with rubyobj.Timeout.Request objects, each 288, which comes to 105MB, 82% of the 128MB heap:
@headius thank you for the detailed analysis! > The example is an extreme case, creating millions of timeout requests per second. To provide more light into the use case, in...
I've added a comment to the thread about ractor support https://github.com/ruby/timeout/issues/53#issuecomment-2597922761. I don't believe the issue will be addressed until this is actual issue in MRI, which will only happen...
Moved the comment to a new issue https://github.com/ruby/timeout/issues/58
> Here you only need one long to count the number of batches, the number of events we already track I think and just divide right? yep that solves the...