[kwlug-disc] Service resource management: i.e. Mailman3 vs Mailman2

[Ronald Barnes]

Khalid Baheyeldin wrote on 2025-01-31 19:37:

> Try something like this:
> 
>   ps -eo pid,user,cputimes,rss,args | grep pyth

I find the output unintelligible .


Modifying it to this:

(ps -eo pid,user,cputimes,rss,args | head -n 1; ps -eo 
pid,user,cputimes,rss,args | grep python) | grep -v grep | less

puts a header row on top.


> The 3rd column is cumulative CPU usage, and the 4th is the resident set 
> size (actual memory used)

There are a bunch of other python processes in there (a DNS tool, 
network dispatcher tool, unattended upgrades tool, but now the grep 
itself is gone (MM2).



> See if some processes are significantly larger than the others, whether 
> the bloat is proportional from v2 to v3, or some have grown, others 
> didn't

The RSS columns are roughly 6× higher on a freshly started MM3 vs a 
long-running MM2.

I don't know what that tells me, exactly.



Frankly, this is just a couple of wins in systemd's corner.

Asking for resources on a process doesn't leave me filtering out 
unrelated processes, the sums are done for me, and there can be stats 
about CPU, IO, IP, and more, all in one place without faffing about.


But - appreciate the help Khalid! Thanks.