Re: /proc/self/oom_adj is deprecated in newer Linux kernels

On 2014-06-10 11:35:23 -0400, Robert Haas wrote:
> > Because it would convert the intended behavior (postmaster and only
> > postmaster is exempt from OOM kill) into a situation where possibly
> > all of the database processes are exempt from OOM kill, at the whim
> > of somebody who should not have the privilege to decide that.
> 
> Gurjeet already refused that argument.

Only that he was wrong:
~# echo -500 > /proc/$BASHPID/oom_score_adj
~# su andres
~$ echo -1000 > /proc/$BASHPID/oom_score_adj
bash: echo: write error: Permission denied # so I can't decrease the likelihood
~$ echo -400 > /proc/$BASHPID/oom_score_adj # but I can increase it
~$ echo -500 > /proc/$BASHPID/oom_score_adj # but also *reset* it

Since we have to do the adjustment after the fork - otherwise the
postmaster would be vulnerable for a time - normal backends can reset
their score.

> Apparently, the child
> processes can only increase their chance of being OOM-killed, not
> decrease it, so you have this exactly backwards: right now, an
> individual system administrator can exempt all of the database
> processes from OOM kill, but cannot exempt the postmaster alone.

Well, if you compile postgres with the #define it'll reset the
likelihood after the fork? That's the reset? Or do you mean without
compiling postgres with the option?

Greetings,

Andres Freund

-- Andres Freund                       http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training &
Services