Filebeat 5.0.0-alpha5 multiple harvesters for same file sending the same messages over and over

Garbage turns out to be coming from our developers...

Coincidental with my trying to roll out the new filebeat.

Looking good so far. I will know better tomorrow.

:slight_smile: Keep me posted.

@viveklak @Tim_Burt A big thank you to you two for keep investigating this issue and pushing forward to get it fixed. I really appreciate all the work you put into this one.

Just got another one.. But perhaps more useful info.

[root@ip-172-16-59-236 rails]# filebeat -version
filebeat version 6.0.0-alpha1 (amd64), libbeat 6.0.0-alpha1

Looks like 6 million identical records in 2 ms.

Here is the caveat.. I have a developer that is dumping a large JSON object into the rails log files. This single line is approx 118 K in length. These lines are prevalent in the rails logs

The line that is duplicated in the Kibana screenshot is surrounded by these long lines.

Could it be possible this could be a buffer overrun?

I can share with you the raw logs but not in this forum. You can contact me at: removed

@Tim_Burt Just sent you an email, interested to see the log file. Can we reopen a new thread for this topic? I get the impression it is also an issue but not directly related to the previous one we that got fixed.

Thanks for the log file. I tried to reproduce the issue locally but couldn't so far. I tested on my side with File output instead of Kafka to take Kafka out of the equation.

Do you see this happen every time such a JSON thingy is in the log or only from time to time? Can you reproduce it locally on your side with Kafka?

One idea could be that Kafka only accepts a certain message size?

Thanks for the effort... Yes, I am aware that my issues may be with Kafka, it is difficult to discern which element is misbehaving. I do have a test environment for file input, but I have not had the time at the moment. We just did a migration and we are all still scrambling with the mop-up.

Your fix seems to have done the trick. I am no longer getting the millions of duplicates. i have seen 2 or 4 duplicates of a line, but that may have a different origin entirely. I will need to do much more research.

Thank you for all of your help and hard work. We are looking forward to 5.0.0.

I think we can close this topic for now.

Please ping us when you hit issue again (also with Kafka :wink: ).

Side note: Beta1 was just released: https://www.elastic.co/blog/elastic-stack-release-5-0-0-beta1

Ping.... I sent you an email with logs and configs....

Nothing I cannot handle, but it may help you solve a corner case.

Thanks again for your help!

This topic was automatically closed after 21 days. New replies are no longer allowed.

I will look into it as soon as I get the time and will get back to you.

For everyone going through this Thread, we found the potential problem here: Filebeat 5.0beta1 with kafka output: multiplication of log lines