Time  Nick     Message
14:32 Pax      so what are you looking for a an invetory software? I'm wondering if a front end for exported resources + location/rack data might do it
14:34 pdurbin  well, shuff and whorka know what i'm used to. . . a custom mysql database with the inventory information we cared about
14:35 pdurbin  which we populated both programatically and manually
14:36 pdurbin  programattically whenever we could, of course.  but a server doesn't know what rack it's in, so that was (is for whorka) manual
14:42 pdurbin  i hadn't heard (or forgot) about exported resources, but i found this: http://docs.puppetlabs.com/guides/exported_resources.html
14:42 shuff    yup
14:44 pdurbin  where do you guys store the information of where a server is physically? (which data center, which rack, which rack unit)
14:44 shuff    us? racktables
14:47 pdurbin  yeah, we use racktables too, but racktables could use a RESTful API: https://projects.cs50.net/project/97/build-a-restful-api-for-racktables-at-philip-durbin/
14:54 Pax      so I was  thinking maybe a custom fact based on CDP to get location
14:54 pdurbin  CDP?
14:54 Pax      all the network devices have 6digit Harvard Location codes that we can get out of snmp
14:55 Pax      Cisco Discovery Protocol
14:55 Pax      it lets the devices know what device they are "next to" on the network
14:55 pdurbin  not all of our network equipment is cisco.  some of it is dell
14:56 Pax      I suspect CPD would still work though
14:59 pdurbin  looking at http://linuxdynasty.org/82/howto-get-cdp-neighbor-information-through-python-and-snmp/
14:59 Pax      but yeah, my though was if we could get the servers picking up location, and ideally rack location that would rule
15:06 pdurbin  Pax: this is certainly a technique i hadn't considered. thanks
15:24 Pax      hmm I like the snap query!
15:26 pdurbin  Pax: snap query?
16:29 Pax      snmp
16:30 Pax      someday I'll learn to type
16:32 Pax      just popped this up http://personalskynet.blogspot.com/2012/01/monitoring-with-puppet-and-nagios.html
16:33 Pax      I'm still kinda getting my bloggy feet under me, but I figured it might help someone else :)
16:34 shuff    huh
16:34 shuff    the meat of that topic seems to be here: http://blog.gurski.org/index.php/2010/01/28/automatic-monitoring-with-puppet-and-nagios/
16:38 agperson so you need the storedconfigs to do the collections?
16:38 shuff    yes
16:39 shuff    puppet agents spit out their information, and it goes in the storedconfig
16:39 shuff    then, when the master is building the catalog for the nagios server, it reads the stored configs and generates a bunch of nagios resources based on them
16:39 shuff    (if i'm interpreting this setup correctly)
16:39 agperson i'm wondering if that means that nagios is always one run behind
16:41 agperson alternately you can setup Zabbix instead and do automatic checkin :)
16:41 shuff    ?
16:41 shuff    why would it be one run behind?
16:41 agperson if the nagios resources are being collected from the stored information, and the stored information is coming from the run reports from the agents, wouldn't that mean that the info is slightly "stale" by its nature?
16:42 agperson probably not enough to be an issue in most cases, but its not like a machine is updated with a new service and instantly nagios is updated with a monitor for that service
16:43 shuff    aha, i see
16:44 shuff    hm
16:45 shuff    to be precise: it looks like the process that generates exported resources is the *compilation* of a given node's catalog, which takes place on the master
16:46 shuff    and since that's the same process that pushes new packages, services etc. to the managed host, i don't believe that this process introduces any lag into the generated nagios config
16:46 shuff    other than the timelag between when the puppet agent runs on the nagios server and on the other server
16:47 agperson got it.  well, it would certainly reduce time lag around here, where manually getting checks entered can take days, weeks, or we can forget to do it entirely! :)
16:49 shuff    srsly :)
18:49 pdurbin  cobbler + git = awesome: scm_track_enabled docs?: https://fedorahosted.org/pipermail/cobbler/2009-June/004301.html
18:56 pdurbin  https://github.com/cobbler/cobbler/blob/master/cobbler/modules/scm_track.py
18:58 shuff    booooo utils.subprocess_call
18:58 shuff    that is heinous
18:58 shuff    no wonder he's complaining that it's slow on every operation
18:59 pdurbin  at least i can easily see which git commands he's running :)
18:59 shuff    he should be using Dulwich (http://www.samba.org/~jelmer/dulwich/)
19:01 shuff    although in general this seems like a cool idea
19:01 shuff    i take it that /var/lib/cobbler is essentially the DB in cobbler?
19:24 pdurbin  that's where most of the configs are stored
19:24 pdurbin  though i do find myself editing /etc/cobbler/dhcp.template sometimes