Memory leak? (swap_pager: out of swap space)

[mpfusion]

There seems to be a memory leak somewhere resulting in the GUI being killed.

The GUI can be relaunched with “/usr/local/etc/rc.d/django restart”. The system is not running any plugins or services except for SSH.

How to fix this or determine what exactly consumes all the memory?

I have two systems showing the issue, both run FreeNAS-9.2.1.3-RELEASE-x64

 

[jgreco]

It sure looks like a classic memory leak.

Without knowing what is breaking, it is hard to advise. If you can catch it during one of these periods, you can run "top" to examine what processes are running and how much memory they are consuming.

I believe FreeBSD still employs a process-killing mechanism that favors hitting the piggy process, so it seems likely that something bad was happening in python.

 

[mpfusion]

If you can catch it during one of these periods, you can run "top" to examine what processes are running and how much memory they are consuming.

I set up a cron job which logs the most memory consuming programs to a file. When the GUI died next time, I'll check if it's pythor or some other process.

I believe FreeBSD still employs a process-killing mechanism that favors hitting the piggy process, so it seems likely that something bad was happening in python.

That would make sense.

 

[mpfusion]

Which column is the correct one to monitor, “SIZE” or “RES”?

 

[jgreco]

SIZE. "RES" is resident, in other words, how much of SIZE is actually currently residing in core.

 

[mpfusion]

The GUI died again today. The problematic task seems to be indeed “python2.7”, shortly before it was killed it used 30079M (SIZE column of “top”).

 

[mpfusion]

I created ticket #4759.