0

I'm a running ubuntu 12.04 LTS server edition and two times in the pasts 48 hours or so the VPS becomes completely unresponsive. Nginx stops responding to requests. There is log activity sent to my logging service. New Relic data stops being sent. I can't even ssh into the box. The only way for me to fix the issue is to go my hosting provider's management system and do a hard restart of the VPS. After the restart everything seems fine until it happens again.

I have 1 GB of RAM plus at least another 1 GB in swap.

I looked though the logs and I have found no evidence of OOM killing any processes.

Below are some screenshots from NewRelic about the server load.

Load average

CPU usage

RAM Usage

Is there any other place I should look for details? Could be kernel bug.

Mab879
  • 103
  • 4

2 Answers2

1

If it's VPS, OOM message usually appears on the hardware node. Check with your hosting provider and see if they can find any OOM issues related to your VPS.

Prasad
  • 11
  • 2
  • This is good advice, not just specifically, but generally also: your VPS provider should be one of your first ports of call, because only they can tell you about effects from outside your machine. – MadHatter Feb 27 '14 at 10:19
0

Generally in case of hung scenario log doesn't help much as syslog/rsyslog process also hung and stop writing logs.To deal with these kind of situation please configure kdump along with sysrq and generate vmcore.This will give you what is actually going on when this issue occur

I don't know much about ubuntu but I found this link https://wiki.ubuntu.com/Kernel/CrashdumpRecipe

How to use crash utility http://people.redhat.com/anderson/

Prashant Lakhera
  • 713
  • 2
  • 10
  • 25