16.0.4 intermittent shudown occuring on vps server

[emonti]

Hi,.

I'm running Ubuntu 16.04 on a remote VPS server.

The problem is that the server is shutting down on its own occasionally. This is a problem because I have a live site which is occasionally highly relied upon to convey schedule data. So, I'm trying to find out what is causing the VPS ubuntu node instance to shutdown.

I tried putting in a K99 script in /etc/init.d(linked to in rc6.d), to output some relevant data like currently running processes, but it seems the script is not being executed.

I don't know if the 2 problems are related, hence why I'm mentioning that.

I thnik it might have something to do with the unattended-upgrade but can't be sure because of the random timing element (hard to track).

I can see no clues in the logs (syslog, dmesg) other than an apparent memory leak i.e.

[   1594]   113  1594   254622    39989   647168        0             0 mysqld
[12619237.856117] oom-kill:constraint [...]
[12619237.856162] Memory cgroup out of memory: Killed process 1594 (mysqld) total-vm:1018488kB, anon-rss:159956kB, file-rss:0kB, shmem-rss:0kB, UID:113 pgtables:632kB oom_score_adj:0
[12619237.874169] oom_reaper: reaped process 1594 (mysqld), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
[12619268.507240] mysqld invoked oom-killer:

'mysqld' occurrs in the above dmesg log a similar number of times as the number of auto shutdowns I have experienced and subsequently had to get VPS support guy to start my server.

As always, any help would be much appreciated.

p.s. only one of these memory leaks is associated with unattended-upgrade.

[TheFu]

16.04 support ends this month.
Not really worth figuring out anything. Move to the next LTS. 18.04 or 20.04 ASAP.

If the system runs out of memory, it will crash. You can delay it by increasing the swap size or adding additional swap.

16.04 is systemd-based, not init, so there aren't any init.d/ scripts fully supported.

But really, the correct answer is to move off 16.04. If you cannot, buy the ESM from Canonical to get security patches for a year, as you plan the migration. https://ubuntu.com/security/esm

[emonti]

Okay, thanks for the advice.

[TheFu]

Okay, thanks for the advice.

BTW, you aren't alone. I have 1 remaining server still running 16.04. This weekend, we have extended downtime scheduled to perform the OS upgrade and multiple upgrades for the software that system runs. I've been testing upgrades in virtual machines the last few weeks and have seen some interesting failures. Hopefully, early on Saturday morning, all will go smooth.