| ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds |
| 02:59 rhce7320 | I am building edubuntu-desktop fai clients. How do I populate /etc/hostname with the reverse lookup of its ip address, or its 'correct' hostname? At the moment all my test clients get the hostname 'host' & it is crippled until I change it to its dns name, eg, fai-131. |
| 03:00 rhce7320 | I have to go back to work now. Will see any responses when I get back in a few hours. |
| rhce7320 (rhce7320@203.57.80.22) left #fai. |
| rhce7320 (~rhce7320@203.57.80.22) joined #fai. |
| lazyb0y (~henning@faiwiki.informatik.uni-koeln.de) left irc: Remote host closed the connection |
| lazyb0y (~henning@faiwiki.informatik.uni-koeln.de) joined #fai. |
| codice (0q4clL5Njg@71.92.66.39) left irc: Ping timeout: 480 seconds |
| 07:24 stockholm | rhce7320: what does your "script" look like that populates hostname? |
| 07:24 stockholm | mine looks like this: |
| 07:25 stockholm | 1693 X Nov 25 Jeremie Miller (0.4K) |
| 07:25 stockholm | err |
| 07:25 stockholm | hostname -f > "$2" |
| 07:26 stockholm | its in the DEFAULT class |
| 07:27 stockholm | i am guessing you have the same? |
| 07:27 stockholm | there are many ways to achive the same |
| 07:27 maswan | we echo $HOSTNAME > $target/etc/hostname |
| 07:31 stockholm | actually we define stuff like LONG_HOSTNAME, SHORT_HOSTNAME, PUBLIC_NETWORKS, SERVICE_NETWORK_IP etc in the environment that we insert in preinst like this |
| 07:32 stockholm | perl -p -i -e 's/##([^#]+)##/$ENV{$1}/g' $2· |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Remote host closed the connection |
| 07:34 stockholm | that way you just need to know your environment vars by heart and can use the same preinst everywhere |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. |
| codice (hus8IDp3td@71.92.66.39) joined #fai. |
| ErKa (keryell@m13.wifi.enstb.org) joined #fai. |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving |
| Mrfai (~lange@suenner.informatik.uni-koeln.de) joined #fai. |
| mettler (~mettler@p578b6607.dip0.t-ipconnect.de) joined #fai. |
| 12:01 mettler | moin moin |
| 12:04 mettler | is there a way to get into Busybox directly ? |
| 12:04 mettler | ..having problems to obtain an ip from dhcp server (in a vlan) |
| 12:26 Mrfai | you can add the option debug to the kernel command line |
| 12:26 Mrfai | I think you will enter the busybox shell, if you have wrong nfs parameters |
| 12:28 Mrfai | http://www.informatik.uni-koeln.de/fai/fai-guide/ch-booting.html#s-booterror |
| 12:29 Mrfai | Seems that you get the busybox shell if the mount does not work at all |
| 12:34 mettler | ok thx! |
| 12:41 mettler | hmm, das prob ist: ich bin in einem VLAN netz. Eigentlich sollte nichts funktionieren was nicht getaggt ist. |
| 12:42 mettler | warum auch immer klappt PXE wunderbar (bootp server wird vom switch getaggt) |
| 12:42 mettler | also ich empfange kernel+initrd, aber kriege dann keine NFS_mounts zum FAI server durch |
| 12:43 mettler | hatte das eigentlich so gemacht, dass ich in der INITRD einfahch ein "vconfig add eth0 12" eingebaut hab. |
| 12:44 mettler | und dann halt "ipconfig ${DEVICE} | tee /netboot.config" |
| 12:44 mettler | als kernelParam: netdevice=eth0.12 |
| 12:45 mettler | leider bleibt der client dann nach der initrd haengen und wartet auf das Mounten des NFSs...was leier nicht klappt |
| 12:45 mettler | hat jmd. vllt. einen generellen tipp, wie ich das besser in einem VLan machen kann? |
| 12:46 mettler | der debug parameter hilft mir leider an der stelle nicht weiter ...und die BusyBox kommt leider auch nicht |
| 12:48 Mrfai | versuchs mal irgendein unsinnigen parameter. Vielleicht root=/dev/xxx statt /dev/nfs |
| 12:53 mettler | ach danke!!! 'getting busy in my box' -das hilft schon mal weiter ;) |
| 12:58 Mrfai | ah es gibt eine option break. Versuch die mal bitte |
| 12:58 Mrfai | anscheinend stopt die bei premount |
| 12:59 mettler | ah, sehr gut (konnte grad aus der busybox "fast" NFS mounten -requests kommen an) ...break |
| 13:02 Mrfai | d.h break klapt? |
| 13:03 mettler | ne, mom |
| 13:03 mettler | hab hier nur eine bladeConsole ....dauert alles ewig |
| 13:04 mettler | grad war nach der break option die tastatur weg. kann aber nicht sagen ob das nicht vllt. an der KVM lag |
| 13:04 mettler | also break bricht auf jeden fall ab |
| 13:04 mettler | aber ich kann leider nichts tippen/scrollenm |
| 13:05 mettler | aber hab jetzt die Initrd neu gebaut -und wenn alles klappt .....:D |
| 13:07 mettler | hmm schade, kernel panic ;) |
| 13:07 oz_ | schön zu sehn wie das wer mit humor nimmt |
| 13:09 mettler | ist erst 2tes lehrjahr und die blades sind alle gefai'd ^^ ..wenn die jetzt am netzwerk was aendern krieg ich halt erstmal panic ;) |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) joined #fai. |
| 13:28 mettler | danke freunde! es bootet wieder ;) |
| 13:30 mettler | das device war scheinbar noch nicht oben, als ich mit vconfig ein vlan adden wollte |
| yournick (~yournick@pD95F7661.dip.t-dialin.net) joined #fai. |
| 14:41 Mrfai | mettler: was fuer eine Lehre machst du? Wo? Was macht ihr mit euren Blades? |
| 14:46 mettler | ich mach Fisi, also fachinformatiker SysIntegration bei der gmit-gmbh.de -die maschinen sind von einem Kunden und dienen als DVB-H plattform |
| 14:46 mettler | hab eine email fuer euch schon ausgefuellt ;) also dieses questionair -muss das nur eben absegnen lassen und schick es euch ;) |
| 14:47 Mrfai | super, das wollt ich hoeren :-) |
| 14:47 mettler | sollte einen netzInstaller aufsetzen und bin *jeha* beim FAI gelandet :) |
| 14:48 oz_ | beste landung, mein herr. |
| mib_co6tf4 (41d66627@webchat.mibbit.com) joined #fai. |
| ErKa (keryell@m13.wifi.enstb.org) left irc: Ping timeout: 480 seconds |
| rgsteele||work (~ryan@75.147.74.137) joined #fai. |
| 15:26 rgsteele||work | Is there a way I can declare a variable in the main FAI process, so that I can export it to every script and config file that needs it? |
| 15:27 mib_co6tf4 | like an environment var? |
| 15:27 rgsteele||work | Ah, perhaps I can declare it in DEFAULT.var |
| 15:28 rgsteele||work | Yep, section 7.6 of the FAI guide confirms. |
| 15:28 rgsteele||work | Woot. |
| 15:28 mib_co6tf4 | excellent. my recommendation was going to be to look at Puppet if you did not find a better solution |
| 15:28 mib_co6tf4 | anybody using FAI as a component in an automation stack? |
| 15:28 rgsteele||work | mib_co6tf4: I'm going to use Puppet, but mostly as a post-installation config mgmt system. |
| 15:28 rgsteele||work | Instead of doing FAI softupdates. |
| 15:28 mib_co6tf4 | ueaj tjat |
| 15:29 rgsteele||work | Slide your fingers to the left one key :) |
| 15:29 mib_co6tf4 | errrk keyboard. sry. yeah that's what i'm planning too... still haven't actually done it yet though so i'm not sure where all the pieces fit together |
| 15:29 mib_co6tf4 | :) what kind of infrastructure are you building? |
| 15:30 mib_co6tf4 | (ie supporting 1 app, many apps, virtualization... ) |
| 15:30 rgsteele||work | High availability database/web/monitoring/mail/etc. clusters. |
| 15:30 rgsteele||work | Virtualization and a flagship app in the mix too. |
| 15:31 mib_co6tf4 | nice! |
| 15:31 mib_co6tf4 | that's similar to what i'm trying to do |
| 15:31 rgsteele||work | Just rolled out a FAI database cluster the other day. |
| 15:31 rgsteele||work | First official FAI rollout, so it obviously took longer than (hopefully) the rest will. |
| 15:32 mib_co6tf4 | of course... did you set up an apt mirror locally? |
| 15:32 rgsteele||work | Yep. |
| 15:32 rgsteele||work | 8 db nodes in 11 minutes. |
| 15:32 mib_co6tf4 | with debmirror or fai mirror |
| 15:32 rgsteele||work | debmirror |
| 15:32 mib_co6tf4 | holy smokes! that's awesome |
| 15:32 mib_co6tf4 | ok so you have the full package repos? |
| 15:32 rgsteele||work | Aye. |
| 15:32 mib_co6tf4 | i was going to try and just get the packages i use |
| 15:32 mib_co6tf4 | with fai-mirror... |
| 15:33 rgsteele||work | mib_co6tf4: Never looked at fai-mirror myself. But I use a local repository anyways, before I set up FAI. |
| 15:33 mib_co6tf4 | my plan is for a HA virtualization cluster that supports Xen (for Windows VMs) and OpenVZ (linux ve's) |
| 15:33 rgsteele||work | I just sync that every night. |
| 15:33 mib_co6tf4 | hmm. I am short on storage space, but that's a good way to do it |
| 15:33 mib_co6tf4 | but maybe I'm looking at this wrong. |
| 15:34 mib_co6tf4 | maybe i need the Virt cluster for app servers, but an HA db cluster for the backends |
| 15:34 mib_co6tf4 | may get better db performance that way... do you run MySQL or Postgres (or both, and if so do they play nice on the same cluster do you know?) |
| 15:35 rgsteele||work | Postgres - it's far and away a better database. |
| 15:35 rgsteele||work | Once your databases start to increase past a couple hundred MB's, PG's performance blows MySQL out of the water. |
| 15:35 mib_co6tf4 | even for lightweight stuff? transactional systems and whatnot? i've been led to believe mysql is faster |
| 15:36 rgsteele||work | Maybe if you're dealing with smaller databases. |
| 15:36 rgsteele||work | My db's are hundreds and hundreds of MB, MySQL just wasn't an option. |
| 15:36 mib_co6tf4 | i'm in a service company, i was thinking mysql for the txn systems, then pull that data out for fulfillment and into PGsql or Bizgres for BI |
| 15:36 mib_co6tf4 | ... regardless though, I wonder if I shouldn't run the DB server(s) on bare metal |
| 15:38 rgsteele||work | Our DB nodes all run Ubuntu Server, and the only packages we install are the ones relevant to Postgres. |
| 15:38 rgsteele||work | Our transaction load is extremely heavy, but we've been very satisfied with the performance. |
| 15:39 mib_co6tf4 | what's your virtualization strategy, if you don't mind my asking... i've just recently brought linux to where i work and, virtualization is new here too... we have 3 esxi servers and they're unwieldy to manage, so i'm looking for options... how to manage all VMs/VEs from a single place, and have the load distributed over multiple nodes, with HA/failover |
| 15:39 mib_co6tf4 | hmm. I wonder how big of a performance hit i'll take running the DB's in VMs. |
| 15:39 rgsteele||work | mib_co6tf4: We really only virtualize our testing servers, and for that we use VMWare Enterprise. |
| 15:39 rgsteele||work | For our production nodes, it's bare metal. |
| 15:40 rgsteele||work | At least for the time being, and no plans to change in the near future. |
| 15:40 mib_co6tf4 | our goal is to be a cloud provider within our company/industry |
| 15:40 rgsteele||work | Cloud computing is an interesting concept that others here are dabbling with, but I've yet to put my hand in to that cookie jar. |
| 15:40 mib_co6tf4 | so our clients have to be able to create/admin their db's from a console or something without having access to any of the other data |
| 15:41 mib_co6tf4 | i've got my hand in too many cookie jars... it's all such cool stuff but i need to focus on one aspect and just get it done... hence, here i am, trying to not set up debian/ubuntu by hand anymore. |
| 15:41 rgsteele||work | Can't say I have much advice for you there, having little experience with cloud computing. I only know what I've read. |
| 15:41 mib_co6tf4 | i figure that's step one, cause vm or bare metal, it's gotta have an OS. |
| 15:42 rgsteele||work | This will definitely help decrease your overhead, once you get it up and running. Anyways, I gotta get back to doing what they pay me for :) |
| 15:42 mib_co6tf4 | your DB cluster, if one drive fails in a node, do you know? if a node goes down, what's the impact? |
| archie (~aa@i-83-67-49-2.freedom2surf.net) left irc: Ping timeout: 480 seconds |
| 15:42 mib_co6tf4 | oh, ok... well thanks for sharing the info! it's given me some things to consider. good luck with your variables :) |
| 15:43 mib_co6tf4 | if you have any links or anything to share about HA capability/failover in your DB cluster that would really be hugely helpful... happy thanksgiving! |
| 15:43 rgsteele||work | We have hardware RAID in the individual nodes, and our databases are sharded (look it up on Wikipedia) |
| 15:44 rgsteele||work | I get notified via email if a drive goes south, so I can go down and swap in a good one (all of 'em are hot swappable) |
| 15:44 mib_co6tf4 | what's the mechanism for sending that email? shell script? |
| 15:44 mib_co6tf4 | it sounds like you have done many of the things i'm working on doing... |
| 15:44 rgsteele||work | If a node goes down, there is some service impact (since they're sharded, and not completely redundant outside of the hardware RAID). |
| 15:45 mib_co6tf4 | ok so not a HA cluster |
| 15:45 rgsteele||work | But, the cost/benefit ratio just doesn't add up when considering 100% redundancy for every single db node (there are hundreds...). |
| 15:45 rgsteele||work | mib_co6tf4: The db cluster isn't. The web cluster/monitoring cluster are. |
| 15:45 mib_co6tf4 | sounds like you run/work in a pretty awesome DC :) |
| 15:45 rgsteele||work | It's not a DC :) |
| 15:46 rgsteele||work | Private company, one flagship app, no red tape or corporate bullshit. |
| 15:46 mib_co6tf4 | hundreds of db nodes, clusters... how is that not? |
| 15:46 rgsteele||work | We have cages at several colo's. |
| 15:46 rgsteele||work | Big cages :) |
| 15:46 mib_co6tf4 | lucky. |
| 15:46 mib_co6tf4 | either way, it sounds really cool |
| 15:47 rgsteele||work | Yeah, I dig it. |
| 15:47 mib_co6tf4 | ah, oh yeah, so how does the server know to email you? |
| 15:47 mib_co6tf4 | that's a critical piece i'll need... if you have a sec... |
| 15:47 rgsteele||work | Nagios. |
| 15:47 mib_co6tf4 | (don't want you to get in hot water at your sweet gig) otherwise i'll be around later |
| 15:47 rgsteele||work | I monitor *everything* with Nagios. |
| 15:47 mib_co6tf4 | i bet you use iclassify too don't you |
| 15:47 rgsteele||work | Nope. |
| 15:47 mib_co6tf4 | hmm. inventory system? |
| 15:48 rgsteele||work | At least, not yet. Maybe I will once I have more time to play with Puppet. |
| 15:48 mib_co6tf4 | (and now i know you are not adam jacob) |
| 15:48 rgsteele||work | I've been called a lot of things, but Adam is not among them. |
| 15:49 mib_co6tf4 | add awesome cluster guru to the list, have a great day man! |
| 15:49 rgsteele||work | You too :) |
| 15:49 mib_co6tf4 | and if you didn't konw about hjk solutions (you probably do) but maybe worth checking out |
| oz_ (~osburg@svn.ph-freiburg.de) left irc: Ping timeout: 480 seconds |
| oz_ (~osburg@svn.ph-freiburg.de) joined #fai. |
| 16:19 Mrfai | /quit/quit |
| Mrfai (~lange@suenner.informatik.uni-koeln.de) left irc: Quit: leaving |
| 16:24 mettler | /quit/quit ;) |
| mettler (~mettler@p578b6607.dip0.t-ipconnect.de) left irc: Quit: leaving |
| ErKa (keryell@keryell.pck.nerim.net) joined #fai. |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) left irc: Ping timeout: 480 seconds |
| rgsteele||work (~ryan@75.147.74.137) left irc: Remote host closed the connection |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. |
| mib_co6tf4 (41d66627@webchat.mibbit.com) left irc: Quit: http://www.mibbit.com ajax IRC Client |
| fxcp (~ctp@brsg-d9bee48c.pool.mediaWays.net) joined #fai. |
| 22:01 rhce7320 | stockholm: Thanks S, & others for hostname response yesterday. I hav just got back to catch up. |
| 22:04 rhce7320 | Unfortunately, my lab has to be dismantled tonight & packed in a container on Saturday. The metal is headed for a school & mission in Ndola Zambia. I hope to set up a couple of nodes at home to iron out my remaining edubuntu roll-out problems before the container is opened in Ndola in 6 months time. |
| yournick (~yournick@pD95F7661.dip.t-dialin.net) left irc: Read error: Connection reset by peer |
| 22:08 stockholm | rhce7320: exciting |
| 22:09 stockholm | what kind of mission is that? |
| fxcp (~ctp@brsg-d9bee48c.pool.mediaWays.net) left irc: Quit: Ex-Chat |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Ping timeout: 480 seconds |
| ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds |
| --- Thu Nov 27 2008 |