High CPU due to out of memory java

[Martinwiertz]

Hi,

I am experiencing daily unavailability of Zimbra due to a java-process which generates a high load. Daily at approx 14/15hrs my server does not respond via webinterface. When ssh logged in and looking via TOP command I saw once a Clamd-proces. I can't find a schedule of cause why this happens during office hours. I looked for logfiles but can't find it. (see below update from Kernel.log)
Server is dedicated Zimbra. All configs are from Zimbra installation.

Running FOSS9.0 latest patches on Ubuntu18.04

Does anyone know how to solve or investigate?

Update: Kernel.log
Mar 3 14:18:08 www kernel: [597428.612998] nginx invoked oom-killer: gfp_mask=0x14200ca(GFP_HIGHUSER_MOVABLE), nodemask=(null), order=0, oom_score_adj=0
Mar 3 14:18:08 www kernel: [597428.613000] nginx cpuset=/ mems_allowed=0
.
.
.
Mar 3 14:18:08 www kernel: [597428.613322] Out of memory: Kill process 29046 (java) score 262 or sacrifice child
Mar 3 14:18:08 www kernel: [597428.613581] Killed process 29046 (java) total-vm:3803008kB, anon-rss:1089244kB, file-rss:0kB, shmem-rss:0kB
Mar 3 14:18:08 www kernel: [597428.778035] oom_reaper: reaped process 29046 (java), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB

Last logs show first lines: xxx process invoked. This varies (nginx, cron, system-devd) but always ends with kill of java process. Java is Zimbra.

[zimbra-wingnut]

I'm also seeing a new issue with Java on Ubuntu 18.04 and Zimbra 8.8.15 where the CPU spikes to somewhere between 50% and 100% every 45 seconds to a minute and TOP shows Java as the culprit. The issue began a week or so ago but I cannot come up with anything that changed on the system though I suspect it may have started with an update of something. Obviously when I shutdown both Zimbra and Java the issue clears. I'm curious to see how many, if any, others are seeing this.

[Martinwiertz]

Thanks for your reply

Java/Zimbra v9.0 is currently active in my server. Daily in the Kernel.log visible that Java is causing a high load. Yesterday I installed double RAM (total 8Gb). This should be enough for my small, personal, home-user server.

[Martinwiertz]

Problem solved by adding more RAM. Last few days no errors. Added 4Gb RAM.