> 05. Technical Services Committee is looking for Volunteers
> 
>     1) Infra Team
>     The Infra Team is responsible for the ongoing day-to-day operations of
>     LOPSA's infrastructure. This team does things like set up mailing lists, 
>     email accounts/aliases, set up Apache vhosts, respond to monitoring
>     alerts, and handle any tickets that come in from other LOPSA committees.
>     LOPSA's infrastructure is made up of Linux (CentOS), Drupal, CiviCRM,
>     Apache, Mailman, check_mk, and RT.
> 
>     Bonus points: We don't currently use any config management, but absolutely
>     should be. If you have experience retrofitting CM into a legacy
>     environment and want to do it, send us an email!
> 
>     Experience in any of these is awesome. Ideal volunteers have at least 2
>     years experience as a Linux admin. Commitment time is about 2-3 hours a
>     month. You should also be awesome at documentation.
> 

So, I was mulling this LOPSAGram item....

Don't most site deploying a CM, start with a legacy environment?  Or sites
with long running/legacy servers considered an outside case for CM deployments.

Well...we did have huge problems cause by other SAs when we were plunged into
CFEngine. The SA that set it up, had started that a friend at where he was
looking to leave us for told him that CFEngine experience would be a huge
benefit...

And, at first he would break legacy systems, including systems that he
shouldn't be touching (or especially...)  While another SA it was routine for
him to destroy all the systems he was responsible for with it.

Like putting his entire group into the firewall policy class, without check
any of his Solaris servers to get handled correctly.  They all got generic
firewalls - install access only, which existed because he got our datacenter
botted once by jumping a new server and then forgetting about it....so by
default it was open to the world and had telnet, which had a well known
exploit, enabled.....

So, installing cfengine and a first run was part of the completion of a
jumpstart, where the first run does some generic things like setup a very
basic firewall and turn off unnecessary services like telnet.  Though we also
stopped doing jumpstarts in the open range in the datacenter that is outside
of its firewall (and we didn't get a border firewall until a year or two ago.)

Our previous manager had been against such things, because she kept saying it
would destroy everything....last year it nearly did.  A trivial programming
error written by some one that proclaimed he was a god of python programming,
and had friends that backed him on it.  Resulted in every system getting 0
byte passwd files.  Since execution was cron driven....things stayed broken
until we could intervene....suspect cfexecd scheduled would've allowed us to
recover faster/easier....since many systems continued to run, just couldn't
log into them or make changes.  Including a production system for a couple of
months before there were configuration changes we couldn't put off any longer.

There's at least a size check on the copy now....

Meanwhile...when I decided it was time to learn CFengine 3...since the latest
upgrade attempt from 2.x to 3.x seemed imminent....and because I was in the
process of setting up two new servers at home (FreeBSD 9.x based, to replace a
pair of Ubuntu LTS that will be EOL next month).

At first, it was because I wanted something better than the cron/rsync scripts
I was using to synchronize configurations on Ubuntu and work on breaking some
of the bad habits I had picked up from some of my now former colleagues.  As
well as learn other things on my own.

But, turned out to be pretty legacy-ish, as I was starting from scratch with
CFEngine, and had already done quite a big of stuff by hand on the new servers.

Things continue to evolve....and the much more legacy (my Ubuntu servers) has
been much slower going...as those are mainly using CFEngine to address current
issues....like needing to fix permissions for a nagios plugin, everytime the
package is updated (by unattended-upgrades.)

And, I don't interact with these (wired) systems directly as much...so not
finding myself add/modifying things in CFEngine for them.  Where sometimes
I'll blow a whole weekend in CFEngine because I decide that I going to have
CFEngine promise it now and in the future....which has nothing to do with the
projects I had planned to tackle that weekend :)  Sometimes I don't even end
up addressing the problem that I dove in for....

Though I have been thinking about how to manage my Ubuntu laptop with 
CFEngine...

And, I've decided continue with CFEngine 3.x at home, even though as the SA
working on our CFEngine 3 upgrade at work was about to go live....it got the
axe, in favor of Chef.

The initial big reason was that there's no reporting/dashboard in community
edition CFEngine.  So boss decided that we'll go Chef.

Though the Chef is for new servers going forward....where everything under
CFEngine will eventually go away....(which may happen sooner than later....)

Knife was also a factor...since with one group...we had been managing the
servers with CFEngine, and they had been managing their applications with
Puppet...and they've been pushing that they be enabled to become like devops
with their systems.  So we share a Chef server with them, and there's plans to
do the same with one or two other groups.

Though both managers that were behind the Chef/Devops move have left us for
greener jobs elsewhere.

Though there has also been a surprise that we have since purchased Enterprise
Chef.

Now, I'd be lying if I said I wasn't curious about helping with these things
at LOPSA...but unfortunately, I'm barely handling my life now....
_______________________________________________
Discuss mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to