0

I had YouTrack installed on my VPS, and lately it just stopped working.

Installation method I used is described here: https://confluence.jetbrains.com/display/YTD65/YouTrack+JAR+as+a+Service+on+Linux.+Alternative+Method (with the difference that I used Apache2 http server in the end). It was being invoked with command:

exec $JAVA_HOME/bin/java -XX:MaxPermSize=1024m -Djetbrains.youtrack.disableBrowser=true -Djava.awt.headless=true  -Xmx1g -jar $JAR $PORT >> $LOG 2>&1 &

And most important - it has been working very well. Lately it just stopped. I tried to figure out what is going on. Logged output of this java process didn't indicate anything wrong - it looked like process was just disappearing after few seconds from invoking. What is also interesting, output was different each time I started it, I was monitoring it with "top" with my own eyes and it looked like it would sometimes disappear instantly, sometimes after some time, and generally - was being killed every time in different state. I couldn't find any log that would indicate anything wrong.

I have free space on disk and 3GB of RAM on this VPS, 1.6GB is free. After taking up about 400MB of RAM it stops. What baffling me the most is that nothing had being changed on this server.

I tried to look at everything, even changed oracle java 1.8 into openjdk 1.7 but nothing changed. Tried to change YouTrack version to newer (was using 6.0, changed into 6.5) - didn't help. I'm using Ubuntu 14.04 on my server.

I run otrace on this process until it lasted, it's 41MB file, so I will paste here only its last breaths: http://pastebin.com/ddfuc5yn . Any ideas what should I even look at to diagnose the problem? Am I being screwed by VPS provider somehow?

Improve this question

1 Answer 1

Reset to default
0

What kind of virtualization platform is your VPS running? What's happening I think is your process is overloading the server and the admins monitoring the node are killing it. Who's your webhost? :)

Improve this answer
3
  • Virtualization: OpenVZ, mintshost.pl . I will contact them and ask if it is possible that my process is being shut down by them.
    – Łukasz Zaroda
    Jan 26, 2016 at 17:19
  • Exactly what I suspected. It is a common practice. Someone just logs in on the node itself, runs vztop, sees your process using a lot of resources and kills it. This is because OpenVZ doesn't isolate the VMs from each other, so if your VM has a process using a lot of resources, it will affect others. It's either them or OOM killer kicks in and automatically kills your process if the node itself is too overloaded. I usually spotted problematic VMs by running dmesg | grep VZID on the node itself. Good luck.
    – Mugurel
    Jan 26, 2016 at 17:34
  • Yep, that was it - I discussed the matter with them and they fixed the problem on their side :) .
    – Łukasz Zaroda
    Jan 27, 2016 at 10:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .