Need documentation for how is setup


Status Graveyard
Server Operations
7 years ago
3 years ago


(Reporter: Callek, Assigned: jakem)





7 years ago
From chat with IT, it looks like has NO documentation on how it is setup internally, making any triage/dealing with it a huge pain.

Both from my standpoint as a website maintainer, and IT's as the actual maintainers of the host.

We hit this problem tonight of having me push the CVS changes and it taking (so far) almost 3 hours for the main-page to get a related update.

Fox3Mike just found the cron that we use:

fox2mike	Callek: MAILTO=""
	fox2mike	# update
	fox2mike	*/10 0-7,9-23 * * * root (cd /data/build/seamonkeyproject-org ; bin/ 1> /dev/null
	fox2mike	0-10,30-59/10 8 * * * root (cd /data/build/seamonkeyproject-org ; bin/ 1> /dev/null
	fox2mike	20 8 * * * root (cd /data/build/seamonkeyproject-org ; bin/ 1> /dev/null

But we need better docs than that, and we should get any docs that result out of this shared with the SeaMonkey team (either as a static-snapshot, or hosted in a public place) -- at least where it makes sense.
All of the scripts involved in building this site are included in your CVS checkout of the site.  They're in the bin directory of your checkout.

The delay in the site going live is because it's behind a caching reverse-proxy server.  If you need it to go live immediately then you need to file a blocker ticket with IT to flush the cache on the proxy server.

That's about all there is to explain.
Look like the proxy VIP for this site is on the Netscaler (ugh, hasn't moved to Zeus yet)
(In reply to Justin Wood (:Callek) from comment #0)
> fox2mike	Callek: MAILTO=""

I take it this is where any output from the crons are sent to. Maybe as a first step we should change this to my address (or a generic one under that forwards to me, if that is possible).

KaiRo, did you get any mails as a result of the failures on Jan 31?

Comment 4

7 years ago
(In reply to Jens Hatlak (:InvisibleSmiley) from comment #3)
> KaiRo, did you get any mails as a result of the failures on Jan 31?

Actually, I just wanted to note that I don't think I ever really received any emails from this crontab, strangely. I'd have been happy if I had, in the past.
Now we should set this to a different address, I believe would be the best one we have for now, we could add Jens to that (it mostly carries stuff related to SeaMonkey builds right now, but it's explicitely set up for all automated info and warnings we get from our infrastructure, so I guess it matches well enough).

Comment 5

7 years ago

The explicit documentation here was for more than just SeaMonkey, but IT as well.... and for more than just what our TTree/bin directory stuff does.

The idea being that we have some sense of HOW this machine is setup/this website itself. Both for internal IT means when I need to poke people for actual error reasons (like there was here, it was not a matter of simple cache being out of date) AND for our own use in having a useful seamonkey-project staging site.

And then we have the overall desire to move the efforts forward to a different RCS than CVS... and to have a better design for our sites infra overall.

BTW, bkero is whom asked me to file this -- due to the IT needs.
"admin node" is mradm02. Cron is in /etc/cron.d/ .

To see what nodes are serving, log into netscaler and run "show vserver bzorg".
Assignee: server-ops → dgherman
My last comment is obsolete now.
Jake, can you write up a quick doc in mana about this?
Assignee: dgherman → nmaul

Comment 8

6 years ago
Just made this page:

I added lots of good info in there, covering the current build/deploy system, contacts, crons, puppet, and (the lack of) external dependencies.

Let me know if I've overlooked anything crucial (mostly for :justdave and :dumitru).
Last Resolved: 6 years ago
Resolution: --- → FIXED
Product: → Graveyard
You need to log in before you can comment on or make changes to this bug.