Possible Memory Leak?

Hi All,

We deployed filebeat to a handful of Windows Server 2012 VMs over the weekend and are seeing very high memory usage on all of the servers. I dug around and found a bunch of other mem leak issues, but none seemed directly applicable.

I made a couple config changes on another server (added the close and clean entries in the below config) and saw a small decrease, but it's still high.

Our config is pretty basic at this point and we haven't done much in the way of configuration.

- type: log
enabled: true
paths:
- C:\Windows\System32\LogFiles**
- C:\inetpub\logs**.log
- C:\inetpub\logs*
*.txt
- D:\logs**.log
- D:\logs*
*.txt
- D:\weblogs**.log
- D:\weblogs*
*.txt
- D:\content*\docs\App_Logs**.log
- D:\content*\docs\App_Logs*
*.txt
close_inactive: 2m
close_removed: true
clean_removed: true
filebeat.config.modules:
path: ${path.config}/modules.d/*.yml
reload.enabled: false
setup.template.settings:
index.number_of_shards: 3
output.logstash:
hosts: ["xxx.xx.xxxx:xxxx"]
processors:
- add_host_metadata: ~
- add_cloud_metadata: ~

I don't see any errors in the logs. Just harvesters going active and inactive. Not sure how many of them is normal. In the search below, it started ~4000 in 3 seconds.

Here's a metrics line if it's relevant.

2019-04-30T11:44:40.401-0500 INFO [monitoring] log/log.go:144 Non-zero metrics in the last 30s {"monitoring": {"metrics": {"beat":{"cpu":{"system":{"ticks":3269250,"time":{"ms":1734}},"total":{"ticks":31018546,"time":{"ms":16952},"value":31018546},"user":{"ticks":27749296,"time":{"ms":15218}}},"handles":{"open":8122},"info":{"ephemeral_id":"cf87ec45-cb91-43d9-bc3a-194a3a91cc42","uptime":{"ms":61890100}},"memstats":{"gc_next":925008144,"memory_alloc":546863192,"memory_total":2679270156384,"rss":6504448}},"filebeat":{"events":{"added":7612,"done":7612},"harvester":{"closed":3297,"open_files":7890,"running":12176,"started":4196}},"libbeat":{"config":{"module":{"running":0}},"output":{"events":{"acked":119,"batches":1,"total":119},"read":{"bytes":36},"write":{"bytes":12254}},"pipeline":{"clients":1,"events":{"active":1,"filtered":7494,"published":119,"total":7613},"queue":{"acked":119}}},"registrar":{"states":{"current":23180,"update":7612},"writes":{"success":165,"total":165}}}}}

That's likely the OS caching files that filebeat is/has accessed. It's pretty common behaviour across any OS/

Thanks for the response!

It seems to be the filebeat.exe process itself that is eating up the memory, not the OS. So I'm not sure if caching would be the issue.

image

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.