[tech] Samson AD

[Matt Johnston]

Hi,

UCC's samba AD was wonky which caused SMTP at UCC to be flakey the past 
few days.

The test for it is (on motsugo or mailfish):
motsugo# sssctl domain-status -o AD.UCC.GU.UWA.EDU.AU
Online status: Online

(that was showing Offline).

Debugging a bit, samson (the AD server) had samba being OOM (out of 
memory). It was because winbindd had a few processes each using nearly a 
gig of memory (that's 600kB for each of UCC's 1500 username, a bit 
high...)

Turns out /var/cache/samba/gencache.tdb was 1.8G. I've deleted that and 
it seems to have autorecreated OK, and winbind is using less memory.

I've also made samba-ad-dc restart on failure:

systemctl edit samba-ad-dc

[Service]
Restart=on-failure

(that gets added to 
/etc/systemd/system/samba-ad-dc.service.d/override.conf)
I guess that can go in the ansible config when I figure how.

Cheers,
Matt