Page MenuHomePhabricator

beta logstash servers run out of disk space
Closed, ResolvedPublicBUG REPORT

Description

What is the problem?

deployment-logstash04, deployment-logstash05 and deployment-logstash06 root partitions are all full.

In particular, deployment-logstash04's root filesystem fills up on a daily basis.

This means logstash on beta is not working.

In the short-term, we should clean them up.

In the long-term, it would be good to make logstash on beta more reliable.

Event Timeline

@thcipriani is there an ETA on this ticket? Would it be possible to prioritize this? It is impacting our testing for SecurePoll which is high priority because of the ongoing board elections.

As a short term stopgap I've cleaned daemon.log manually on deployment-logstash0[456] (same thing done on all)

deployment-logstash04:/var/log# du -shc /var/log/daemon.log*
3.5G    /var/log/daemon.log
8.6G    /var/log/daemon.log.1
141M    /var/log/daemon.log.2.gz
33M     /var/log/daemon.log.3.gz
13G     total
deployment-logstash04:/var/log# > /var/log/daemon.log
deployment-logstash04:/var/log# > /var/log/daemon.log.1

Longer term looks like there is excessive gc logging enabled

our bit here should be good to go. please re-add observability if you require additional help

Hello. I need help with the issue of deployment-logstash04's root filesystem filling up on a daily basis. I asked a question on T233134: logstash-beta.wmflabs.org does not receive any mediawiki events as well. This (T288989) ticket's description aligns with my problem more directly than T233134's.

Today @Majavah turned off deployment-logstash04, deployment-logstash05, and deployment-logstash06.

colewhite claimed this task.
colewhite subscribed.

Boldly resolving because we have migrated to a new cluster serving beta logs.