=== dendrobates is now known as dendro-afk | ||
=== zyga is now known as zyga-afk | ||
=== Hussain is now known as Guest32903 | ||
=== zyga-afk is now known as zyga | ||
=== dendro-afk is now known as dendrobates | ||
hazmat | smoser, ping.. we're seeing some issues booting instances on ec2, with apt-update failing.. was curious if you knew anything regarding http://paste.ubuntu.com/907227/ cloud-init output.. | 14:29 |
---|---|---|
hazmat | looks like a hash mismatch on the update | 14:30 |
smoser | utlemming, ^ | 14:30 |
hazmat | i thought the whole pipeline thing was default enabled | 14:30 |
smoser | someone needs to open an RT on it, hazmat | 14:30 |
smoser | we have to have IS kick that server | 14:30 |
utlemming | hazmat: run "rm -rf /var/lib/apt/lists/*" | 14:30 |
utlemming | hazmat: and then try again | 14:30 |
* smoser is very looking forward to the hope of this all magically being fixed. | 14:30 | |
hazmat | utlemming, on the instance.. we can't per se.. we rely on cloud-init to install ssh keys etc.. but its failing pretty early. | 14:31 |
smoser | cloud-init should still install keys, no ? | 14:31 |
* hazmat checks | 14:32 | |
hazmat | smoser, it does | 14:32 |
hazmat | but manual intervention is a hoser for juju | 14:33 |
hazmat | smoser, is there an RT already? i'd be happy to file it, i'm just fuzzy on the details of whats needed | 14:33 |
hazmat | utlemming, smoser is this issue in relation to the s3 apt repos, or is it just one of the ec2 pkg repos is borked | 14:34 |
* hazmat tries a different a region | 14:34 | |
utlemming | hazmat: its the ec2 pkg repos | 14:34 |
utlemming | hazmat: so what I think is happening, is that the release file is not being updated. When "curl -I" on the Release file, the information seems stale for us-east-1. If you delete the local release file, then it should work. | 14:35 |
utlemming | hazmat: if you can verify that, then I think we'll have enough information for a RT ticket | 14:36 |
utlemming | hazmat: and I'll happily file it | 14:36 |
smoser | utlemming, we dont really have to verify it | 14:36 |
hazmat | utlemming, i just killed the instance, but i should be able to reproduce and verify in 5m | 14:37 |
utlemming | smoser: true, but verification which proves my theory makes it easier for IS to fix it | 14:38 |
smoser | utlemming, check-archive (i added to make it check Sources.gz and Sources.bz2) runs successfully | 14:44 |
smoser | ./check-archive http://us-east-1.ec2.archive.ubuntu.com/ubuntu/ precise | 14:44 |
hazmat | utlemming, confirmed fwiw | 14:47 |
utlemming | hazmat: okay, thank you kindly. | 14:48 |
smoser | hazmat, what did you confirm ? | 14:49 |
smoser | just that removing those files and trying again fixed it ? | 14:49 |
hazmat | smoser, yup | 14:51 |
utlemming | smoser: in looking at the headers between the us-west-2 and us-east-1 mirrors, the HTTP headers indicate that the us-east-1 records is stale (at least when I looked yesterday). So what I think is happening, based on what I saw yesterday is that apt is using the cached release file because it appears that a newer one doesn't exist. | 14:53 |
utlemming | smoseer: I'm rechecking now | 14:53 |
smoser | utlemming, you're correct | 14:55 |
utlemming | smoser: see http://paste.ubuntu.com/907322/ | 14:55 |
smoser | none of the mirrors in us-east-1 has updated since the 27th | 14:55 |
smoser | utlemming, http://paste.ubuntu.com/907324/ | 14:57 |
utlemming | RT filed | 14:59 |
smoser | utlemming, fwiw, this particular issue woudl have been (i think) fixed if we had the source lists in /var/lib/apt/lists | 15:06 |
=== 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 | ||
=== dendrobates is now known as dendro-afk |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!