| rava | greetings | 00:28 |
|---|---|---|
| rava | anyone know why ec2-run-instances seems to break a security group with | 00:28 |
| rava | spaces in it into an array? | 00:28 |
| rava | i'm passing the value in from a shell script with "my sec group" | 00:29 |
| rava | doing it manually via the commandline seems to work, but wrapped in a var in a bash script and ec2-run-instances seems to think it should eval each word in the string as a seperate sec grp | 00:29 |
| erichammond | rava: Sounds like you might have a shell argument expansion issue with your command line. Perhaps not using the right quoting? | 00:33 |
| erichammond | rava: Or there could be a bug in ec2-run-instances command itself. | 00:34 |
| erichammond | oh, it works on your manual command line, so it's probably a quoting problem in your bash script. | 00:34 |
| rava | thought as much | 00:35 |
| erichammond | If you can pastebin your shell script, somebody might be able to glance at it and give feedback. Make sure you remove any private info. | 00:36 |
| rava | sure thing, one sec while i clean it up :) | 00:36 |
| rava | erichammond: http://pastebin.com/tAAkukmE | 00:40 |
| erichammond | rava: Try sticking the word "eval" in front of "ec2-run-instances" | 00:45 |
| erichammond | rava: ec2_out=$(eval ec2-run-instances ... | 00:46 |
| rava | that worked, i keep forgetting about eval... | 00:47 |
| rava | thanks! | 00:48 |
| === dendrobates is now known as dendro-afk | ||
| theTrav | erichammond: thanks, I figured that might be the case... It kind of implies that there's no neat/easy migration between the two platforms | 00:58 |
| theTrav | except maybe making an image on them that runs a virtualbox image :P | 00:59 |
| erichammond | theTrav: If you are building custom images, then you can easily migrate. | 01:02 |
| erichammond | theTrav: Start with the dowloadable image, customize it, and use it for both EC2 and UEC. | 01:25 |
| erichammond | theTrav: You can also snapshot a running instance on EC2 and build the UEC image from a copy of that file system. | 01:25 |
| erichammond | theTrav: In many cases I recommend starting with a standard base image and applying software installation/configuration at run time. | 01:26 |
| erichammond | theTrav: Then you're not copying big, complicated images around, but just small, clear startup instructions. | 01:28 |
| theTrav | erichammond: yeah, one of the guys here recommended I look into Chef | 01:55 |
| theTrav | there's a pretty nasty closed source dependency on the project though with a fairly horrendous installer, so I'm not sure how well I could script that | 01:56 |
| erichammond | theTrav: Sure, those configuration management frameworks are used by many, but simple shell scripts are often all you need, especially if you are configuring from a known starting point (base image). | 01:56 |
| erichammond | theTrav: The installer would end up creating files. Perhaps you can tar up the results and install them on new systems. | 01:57 |
| === dendro-afk is now known as dendrobates | ||
| === dendrobates is now known as dendro-afk | ||
| === dendro-afk is now known as dendrobates | ||
| === dendrobates is now known as dendro-afk | ||
| === dendro-afk is now known as dendrobates | ||
| TeTeT | Daviey: would you know if there's something like MAX_MEM akin to MAX_CORES where a admin can oversubscribe memory on a NC for UEC? | 15:40 |
| TeTeT | Daviey: question just came up in our UEC class | 15:41 |
| Daviey | yes! | 15:46 |
| Daviey | ih.. no | 15:46 |
| Daviey | TeteT. sadly we cannot currently oversubscribe memory. | 15:47 |
| TeTeT | Daviey: thanks for the quick answer! | 15:47 |
| Daviey | UEC.. uses virt.. not containers.. whuch wiuld allow it | 15:48 |
| Daviey | wow... typo fail. | 15:49 |
| TeTeT | kvm with ksm should be possible to oversubscribe memory, not? | 15:51 |
| TeTeT | other question is: can there be multiple sc in a cluster? I guess not, but not sure | 15:51 |
| elasticdog | if I changed a CC's IP address and want to make sure it gets re-registered, what do I do? | 16:15 |
| elasticdog | "euca_conf --list-clusters" shows the old IP address | 16:15 |
| elasticdog | /var/log/eucalyptus/registration.log seems to indicate that it's "Calling cluster cluster01 ..." at the correct (new) IP address | 16:17 |
| elasticdog | looks like manually deregistering/registering works for the CC/SC...now to fix the NC's | 16:24 |
| Kiall | any of the UEC devs about? (or anyone familiar with patching/compiling euca on ubuntu...) - I keep running into some issues when building saying patches dont remove cleanly... | 16:30 |
| Kiall | elasticdog, you might want to test the hell out of it once you're done :) I've never managed to re-number UEC components without hassles.. | 16:31 |
| elasticdog | Kiall: thanks, I will...the whole process seems a bit wonky | 16:55 |
| Kiall | i don't think your really meant to re-number things ;) | 16:55 |
| elasticdog | Kiall: my whole install is a bit of a mess...basically I wanted to change to static IPs after everything was installed | 16:56 |
| Kiall | yup .. that would do it ;) | 16:56 |
| elasticdog | it's also compounded by installing on new mac mini's, and the ethernet driver isn't in 10.04 by default, so I used a USB->Ethernet adapter to install, then compiled the proper kernel module | 16:59 |
| ubuntucloud254 | login | 17:39 |
| === dendrobates is now known as dendro-afk | ||
| === dendro-afk is now known as dendrobates | ||
| === Kiall is now known as Kiall|AFK | ||
| === Kiall|AFK is now known as Kiall | ||
| === dendrobates is now known as dendro-afk | ||
| === dendro-afk is now known as dendrobates | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!