On Wed, 25 Sep 2002, Steven Wagner wrote:
>
> Uhhhhh yeah, rsync, that's the ticket...
>
> [Just a cron script that runs every 15 minutes - losing 15 minutes' worth
> of data isn't going to freak anybody out over here, since we aren't using
> this for accounting purposes, just general health.  After we reach full
> deployment I'll try tweaking the frequency of updates... ]

trouble is, I actually sometimes have folks say to me "Can I find out how
much compute time was used by project X in 1999 between May and October?"
Never mind that the answer will be totally irrellevant to any future
project, they still want to know, because then they will look well
informed.  I can answer that question for some systems as far back as
1997, even before I worked at this job, thanks to an admin who kept
regular, well labeled backups and had a decent sar system setup.  but now
that we have bunches of distributed clusters instead of a few large
centralised sgi machines, this data is really dispersed, and is growing
ever harder to keep tack of and coallate.

> You know, it's a good thing we aren't developing open-source proton packs...

funny you should say that, as one of the clusters I am using ganglia for
is all about atom smashing.

> OpenBSD doesn't build, eh?  How odd... I guess *BSD don't have as much in
> common as I thought.

could be that I've got autoconf screwed up on the open bsd systems.  I
haven't yet delved into it.

> Custom user metrics are actually given a metric hash value of 0 and are
> stored in a different manner.  And don't blame me, it was like this when I
> got here.  I refer you to the SF archives of this list for my numerous
> whiny e-mails about metric handling.

Don'tworry, I'm not blaming anyone, and if it were'nt for you and your
work on the irix port I would still be fumbling about for a
cross-platform solution.

> And Ganglia has worked that way since at least 2.3.x, which is about when I
> started taking an interest in it.  It doesn't "need" to be that way, that's
> just the way it is... for now.

as matt mentioned, I don't really think it is a problem except for a few
solaris metrics, which I'm not using.

> k`ddd
> Man, I wish I could use something that simple to track jobs here.

at the moment there's not hard money based on it, only vague threats,
which is wy I can live with the fudges and approximations that are a
natural result at this stage.  But vague threats are all I need to keep
hardware flowing towards the engineers who need it, so it has been
effective in that sense.

-- 
Ryan Sweet <[EMAIL PROTECTED]>
Atos Origin Engineering Services
http://www.aoes.nl


Reply via email to