| erichammond | erikd: You might find that AMI storage is pretty cheap (cents per image per month). I have over 500 AMIs listed. | 02:45 |
|---|---|---|
| erichammond | erikd: The entire image is uploaded to S3 in the bundle. | 02:45 |
| erichammond | The only dependencies it might have are the kernel (AKI) and ramdisk (ARI). | 02:45 |
| erichammond | If the kernel or ramdisk are deleted by their owners, then you would need to specify different ones when running your image. | 02:46 |
| cagdas | hi all | 08:15 |
| cagdas | how to we bundle images downloaded from ec2-images.ubuntu.com | 08:16 |
| cagdas | i didn't figure it out | 08:16 |
| cagdas | in other words how do we convert these images to AMIs | 08:17 |
| soren | cagdas: You use ec2-bundle-image. | 08:19 |
| soren | cagdas: Let me see if I can find some instructions somewhere. | 08:19 |
| cagdas | thanks soren | 08:19 |
| cagdas | it would be good if this is added to the topic part of this channel | 08:20 |
| soren | cagdas: I'm not sure we have any good docs on the subject. I recommend you look at the documentation for ec2-bundle-image and just ask me if you have specific questions. | 08:33 |
| cagdas | thanks soren | 08:33 |
| cagdas | i think i handled after your comment | 08:33 |
| cagdas | i used ec2-bundle-image | 08:33 |
| soren | Oh, fantastic. | 08:33 |
| cagdas | then used ec2-upload-bundle | 08:33 |
| cagdas | now i have an AMI | 08:34 |
| cagdas | ans I just launched a new instance after registering my AMI | 08:34 |
| cagdas | however, I'm waiting for ssh to be available on there | 08:34 |
| cagdas | it says operation timeout | 08:35 |
| cagdas | is it possible that ubuntu images do not have sshd working by default? | 08:35 |
| cagdas | (i have opened the ports already on amazon ec2 security group) | 08:35 |
| cagdas | it looks like i failed to define kernel-id and ramdisk-id | 08:48 |
| cagdas | where do i define them? | 08:48 |
| cagdas | while bundling the image? | 08:48 |
| cagdas | or while launching an instance? | 08:48 |
| elmoro | Hello to ALL | 08:52 |
| elmoro | i have a problem with ec2-running-instances (eu-west-1) with the error keypair not exist. I have a working key pair but it will not be found | 08:54 |
| elmoro | also in the aws console i have only one existing keypair - when i use the ec2-describe-keypairs then i get my 3 generated keypairs. | 08:55 |
| soren | elmoro: Keypairs are local to regions. Could that explain it? | 10:21 |
| soren | elmoro: I.e. a keypair generated in us-east-1 does not exist in eu-west-1. | 10:22 |
| elmoro | ok thy | 11:46 |
| elmoro | i think i generated it in US and then later i want to use it in Eu region. Thx | 11:47 |
| deadprogrammer | anyone have trouble with apt-get upgrade of module-init-tools (3.3-pre11-4ubuntu5.8.04.1)? | 15:20 |
| deadprogrammer | sudo apt-get upgrade has me hungup forever on module-init-tools... and my poor server is in quite a sad state | 15:33 |
| === nijaba` is now known as nijaba | ||
| elmoro | anyone has experience with real time kernel for Gameserver ? The ping performance is bad and the cpu is 99% idle? | 16:14 |
| deadprogrammer | sudo apt-get upgrade has me hungup forever on module-init-tools... and my poor server is in quite a sad state :( | 16:19 |
| jeremydei | elmoro: what gameserver are you running? | 17:38 |
| elmoro | hl2 dedicated server | 18:15 |
| elmoro | and in the future cod4 and tf2/battlefield | 18:16 |
| elmoro | but with the small instance - the ping performance is not the best. At the weekend i will try the bigger machines | 18:17 |
| jeremydei | elmoro: interesting. I haven't tried it (running source server) yet, but I would like to. Are you using Elastic IP? you may want to try that also, maybe there's some magic there that can help the ping | 18:23 |
| elmoro | the perfomance seem to be okay (tick 66) | 18:32 |
| elmoro | but the ping looks bad | 18:32 |
| elmoro | yes i use elastic ip.. | 18:32 |
| deadprogrammer | anyone using ebs with xfs? | 18:45 |
| === erichammond1 is now known as erichammond | ||
| === natea is now known as natea|lunch | ||
| === natea|lunch is now known as natea | ||
| erichammond | deadprogrammer: Yes, I use EBS with XFS and MySQL on production servers. | 20:18 |
| deadprogrammer | Hi erichammond | 20:58 |
| deadprogrammer | yes I am running mysql well... this time it is postgresql | 20:59 |
| deadprogrammer | I am running an somewhat older hardy that is based on one of your older amis | 20:59 |
| deadprogrammer | it has been running in production for a while | 20:59 |
| deadprogrammer | until earlier todau when and apt-get upgrade when bad | 21:00 |
| deadprogrammer | module-init-tools (3.3-pre11-4ubuntu5.8.04.1) hung, and then I have been unable to get the instance back to 'normal' | 21:02 |
| deadprogrammer | so then I decided to just launch a new instance from the AMI | 21:03 |
| deadprogrammer | I noticed the snapshot did not appear to have my actual database... it was as if no data had ever been saved to the EBS volume | 21:04 |
| deadprogrammer | hence my original question... anyone running Postgresql on EBS using XFS should have been the specific question | 21:04 |
| erichammond | deadprogrammer: I don't know what might have happened to cause the upgrade to hang. I am not able to reproduce it myself. | 21:06 |
| erichammond | The "empty EBS volume" is a commonly reported problem. | 21:06 |
| deadprogrammer | urg... first time I have run into it | 21:06 |
| erichammond | However, it always seems to turn out to be user error. | 21:06 |
| erichammond | Common causes include: | 21:06 |
| erichammond | - The volume was not mounted where you thought it was | 21:07 |
| erichammond | - The snapshot was of the wrong volume | 21:07 |
| erichammond | - On mounting the new volume, you ran "mkfs" and overwrote the files | 21:07 |
| deadprogrammer | this server has been running since march... and did not re-run mkfs on this restore attempt | 21:08 |
| erichammond | It can be very tricky to get everything in the right place all the time with EBS volumes and snapshots, especially since there the ids are opaque. | 21:08 |
| deadprogrammer | yes, I keep thinking I must have made an error somewhere | 21:09 |
| erichammond | Is the original volume still attached to the instance? | 21:09 |
| deadprogrammer | yes | 21:09 |
| deadprogrammer | but the files are not there on that volume | 21:09 |
| erichammond | You can connect to the instance? | 21:09 |
| deadprogrammer | there are the original databases | 21:09 |
| deadprogrammer | I am still connected to it | 21:09 |
| deadprogrammer | postgresql was removed as part of the apt-get upgrade, so I cannot just start the server to see if there database is "there"... but postgres is fairly transoprent about certain things | 21:10 |
| deadprogrammer | like there is an actual file on disk that show what databases exist for that server | 21:11 |
| erichammond | I'm confused. You have the original EBS volume attached to the original EC2 instance, and you have mounted the file system on that volume, but it is empty? | 21:11 |
| deadprogrammer | not empty... it does not have the database I expect to be there. I can tell by examing the file system on that volume | 21:12 |
| deadprogrammer | likewise, a new instance with an EBS volume created from any snapshot, has only those same databases... but not my production database | 21:13 |
| erichammond | Ah, I see. You had a database in Postgres, but now you can't find where the files for that database reside. | 21:13 |
| deadprogrammer | exactly | 21:14 |
| deadprogrammer | as if postgres never actually wrote any updates to the file system | 21:14 |
| deadprogrammer | despite the facts that it has been running for months | 21:14 |
| deadprogrammer | and has been updated at least once without event | 21:14 |
| erichammond | I don't know Postgres, but if this were MySQL, I would suspect that the configuration was pointing to database files which were not on the EBS volume or that the EBS volume was not mounted at the time the database was running. | 21:21 |
| erichammond | Have you looked "under" the EBS file system mount point? I.e., umount and look at the mount point directory/ | 21:22 |
| erichammond | ? | 21:22 |
| deadprogrammer | how do I look at the mount point directory? | 21:24 |
| erichammond | Where is the EBS volume mounted? | 21:24 |
| deadprogrammer | I just umount /vol | 21:24 |
| deadprogrammer | it is /dev/sdh | 21:24 |
| erichammond | sudo umount /vol | 21:24 |
| erichammond | ls /vol | 21:24 |
| deadprogrammer | device dev/sdh mapped to /vol | 21:24 |
| deadprogrammer | after I umount I see the correct data! | 21:25 |
| deadprogrammer | so do not umount? | 21:26 |
| deadprogrammer | well not sure if "correct data" but at least I see entry in file that indicates that database existing | 21:28 |
| deadprogrammer | I think you have just saved me from myself... thanks erichammond for your time, I sincerely appreciate your expertise | 21:29 |
| deadprogrammer | one last question however. since I seem to have messed up, but having a directory on /dev/sda1 named /vol | 21:33 |
| deadprogrammer | how can I get my EBS volume mounted | 21:33 |
| deadprogrammer | ? | 21:34 |
| deadprogrammer | just correct fstab and mount -a? | 21:34 |
| erichammond | I can't provide specifics since I don't have access to all the details of your configuration, but you probably need to move the existing (good) contents of /vol aside, mount the EBS volume, copy the good contents onto the volume, snapshot the volume, create a new volume, and test with a new instance, or something to that effect. | 21:38 |
| erichammond | Be very careful as your production database is on ephemeral storage and is not backed up at all. | 21:39 |
| erichammond | If that instance goes down, you've lost it. | 21:39 |
| deadprogrammer | yes I realize that! | 21:40 |
| deadprogrammer | I just added this to my fstab | 21:40 |
| deadprogrammer | hmm | 21:41 |
| erichammond | I've updated http://ec2ebs-mysql.notlong.com/ so that the mount point is created mode "000" to help prevent accidents like this. | 21:52 |
| erichammond | I do that in my own systems, but didn't include it in the original tutorial. | 21:52 |
| deadprogrammer | that is a good idea | 21:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!