=== zul_ is now known as zul === SpamapS_ is now known as SpamapS [17:24] Hi, I have a question about hostnames for nodes in 10.04. My nodes are getting hostnames like "ip-172-19-1-2" but /etc/hosts doesn't list this name. Do I need to enable (or rather, NOT disable) the cloud dns to work around this? [17:38] patrickw, you're seeing bug https://bugs.launchpad.net/ubuntu/+source/eucalyptus/+bug/475354 [17:38] Launchpad bug 475354 in eucalyptus "Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)" [Undecided,Fix committed] [17:38] <_mup_> Bug #475354: Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)

Bug #475354: Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)

Launchpad bug 475354 in eucalyptus "Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)" [Undecided,Fix committed] https://launchpad.net/bugs/475354 [17:39] <_mup_> Bug #475354: Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)

Bug #475354: Hostname not set correctly on UEC cloud due to IP address in local-hostname manifest data (DNS)

wow, _mup_ that was a bit excessive, don't you think ? [17:55] I had researched that bug, but it seemed different, though related. I'm not getting a hostname in the form of an dotted IP, but rather "ip-W-X-Y-Z", which was I guess the resolution of the bug. But my machine doesn't resolve that name to an IP... I guess I have to enable cloud DNS and reconfigure my CC to use it for dns? [17:56] Actually, I seem to be getting "euca-W-X-Y-Z" now that I enabled cloud DNS [18:49] patrickw, and euca-W-X-Y-Z should be reverse resolvealbe, right ? [19:39] I'm attempting to build a private cloud for testing. Haven't had success in starting virtual server instances yet. [19:41] Should eucalyptus-nc-publication be installed and running on the frontend controller? It is not currently. [19:47] smoser, W.X.Y.Z does not reverse resolve. At any rate, I don't have control of our DNS servers and doing all this DNS stuff just so a machine knows it's own IP seems kind of nuts... shouldn't something just stick the name into /etc/hosts automatically? [19:48] patrickw, there are reasons that we dont do that in the instance. [19:48] i have to dig to come up with what that is, but it caused issues. [19:50] smoser, at any rate, I may be able to work around my problem by specifying "localhost" as the host name to my application. I've been messing with the euca DNS but my instances don't get a domainname and I don't need to open any *more* cans of worms at this point. :) [19:50] yeah, dns is scary to poke at. [19:50] patrickw, you can very easily modify /etc/hosts on your own in a user-data script or something [19:52] smoser, I was thinking about that, but avoiding scripting that up and just massaging the app I'm trying to run seems a simpler option for the short run [19:53] though I have been shaking the rust off of my bash skills while working on this project. :) [19:59] patrickw, well, it might look something like this: [19:59] ip=$(wget -qO - "${md}/local-ipv4") ; hn=$(wget -qO - "${md}/local-hostname"); echo "ip=${ip} hn=${hn}"; echo "${ip} ${hn}" | sudo tee -a /etc/hosts [19:59] you pass that into user-data with a header of '#!/bin/sh' and it will probably work (or be close to it, i didn't test) [20:00] oops [20:00] md=http://169.254.169.254/latest/meta-data [20:00] forgot that part [20:17] smoser, damn, thanks, I'll save that for later... I may add plain old $(hostname) to the final echo also, as what you did apparently is picking up "eucalyptus.internal" as the domain. [20:18] fwiw, you can dump all metadata with this nice one-liner: [20:18] python -c 'import boto.utils, pprint; pprint.pprint(boto.utils.get_instance_metadata())' [20:20] I think my problem is that I don't have any storage volumes configured. I'll be following https://help.ubuntu.com/community/UEC/StorageController to see if I can get this working. [20:28] smoser, awesome, that will come in very handy. I probably would have tried poking around in /proc or trying to hack the address out of ifconfig... brand new to euca, that web interface will be useful [20:41] Matthew, you don't need to use a storage controller just to get VMs running [20:52] patrickw: I wonder what I'm doing wrong, then. What log files would I look at to determine why VMs are terminating? [21:05] MatthewM: there is some info on this page http://open.eucalyptus.com/wiki/EucalyptusTroubleshooting_v1.6 [21:05] patrickw: thanks === ]oestewart is now known as joestewart|afk === A-Tuin_ is now known as A-Tuin [23:02] getting frustrated. I'm following the documentation to the letter, but instances still terminate. Got one to actually show "running" for about 20 seconds. I got excited, but then it termintated too. [23:02] I'm not sure what to try next. [23:03] Did anyone else have this problem when starting out? [23:37] MatthewM, i did [23:38] MatthewM, i did eventually solve, but i ended up tailing the log files on the compute node and cluster controller, some copious output results.. i think in one case i had a bad image, and in the other my networking wasn't correct [23:39] hazmat: Thanks. I appreciate that. [23:39] hazmat: Do you remember what the networking problem was? The installer should communicate and set up everything automatically, right? [23:41] MatthewM, i didn't have my networking setup properly, and then i tried to switch it from bridge to system, but i didn't realize the cluster controller was keeping cached state, so i needed to perform some command to clear out its instance state and reload them from its config files, this was about 9 months ago, i think the installer has gotten better.. but i haven't really done much with uec/euca since i got it setup to te [23:41] st out. [23:42] the log file output is a bit copious but it did have some pointers [23:42] but i had to tail from both the node controller and the cc to get an idea of the interaction [23:43] MatthewM, you might be able to use ec2-get-console-output to see things from the perspective of inside the instance, if it was successfully running [23:44] hazmat: I only got it to be "running" for a few seconds. Thanks for your suggestions. I'll give it a try again soon. [23:59] Hey cloudy people