| smoser | cornfeedhobo, thanks | 01:00 |
|---|---|---|
| cornfeedhobo | thank you as well :) | 02:32 |
| === cpaelzer is now known as cpaelzer_afk | ||
| jlebon | hi all, hitting a strange issue on f23 with 0.7.6 | 17:00 |
| jlebon | it seems like commands in the bootcmd section are run twice? | 17:00 |
| jlebon | once during init-local and once more during 'init' | 17:01 |
| jlebon | is this the expected behaviour? | 17:02 |
| smoser | jlebon, they should only run at one of those | 17:40 |
| jlebon | smoser: hmm strange. i'll check if fedora is holding any patches that might affect this | 17:42 |
| jlebon | which of the two should it run at, init-local? | 17:42 |
| smoser | well, it should run when the datasource is found. | 17:43 |
| smoser | i'm guessing you're config drive ? | 17:43 |
| jlebon | no, i'm using NoCloud | 17:43 |
| jlebon | giving seedfrom on the commandline | 17:43 |
| smoser | hm. | 17:43 |
| smoser | just thinking, it coudl be a bug. | 17:44 |
| jlebon | hmm, yeah I don't see anything related to this in the fedora patches | 17:45 |
| smoser | jlebon, can you post /var/log/cloud-init* ? | 17:51 |
| smoser | somethign is goign wrong in that if init-local finds a datasource and runs the init modules, then init should not | 17:51 |
| jlebon | smoser: sure, let me get it from a fresh instance | 17:51 |
| smoser | and generally speaking if you're seeding you want 'dsmode=net' . but it shoudl actually do the right thing if it is seeded from http | 17:52 |
| jlebon | smoser: ahhh, i'm using dsmode=nocloud;s=/path/to/cloud-init-data | 17:55 |
| jlebon | smoser: http://ur1.ca/ocj8n | 17:55 |
| jlebon | the ln msg is how I know it's runnign twice | 17:55 |
| jlebon | i didn't use ln -f so it fails on the second time | 17:56 |
| smoser | yeah, so you want: | 17:58 |
| smoser | ds=nocloud,dsmode=nocloud-net | 17:59 |
| jlebon | even though i'm seeding locally? | 17:59 |
| smoser | well, if you want networking to be guaranteed when the bootcmd commands run | 18:00 |
| smoser | then you need dsmode=net | 18:00 |
| smoser | maybe you need journalctl or systemctl info | 18:01 |
| smoser | the pasete there isnt very much other than to show like you said that its probably running twice. although the order there is still odd. | 18:01 |
| smoser | (that they come before the 'running init') | 18:01 |
| smoser | maybe i just misunderstood your problem | 18:01 |
| smoser | bootcmds do run on everyboot | 18:01 |
| smoser | not "per-instance" | 18:01 |
| jlebon | i can paste the full journal if needed | 18:02 |
| jlebon | right, it does run which is good | 18:02 |
| jlebon | my issue is that it runs twice | 18:02 |
| jlebon | it's not a big deal in this case because i can make my bootcmd idempotent | 18:02 |
| jlebon | by using ln -f | 18:02 |
| smoser | it shoudl not run twice per boot though. | 18:02 |
| smoser | it shoudl run once on every boot | 18:03 |
| jlebon | right, so there's a mismatch somewhere -- did you mean earlier that using the ds=nocloud,dsmode=nocloud-net cmdline would work around this issue? | 18:04 |
| jlebon | the files themselves are located locally, so i don't really need net access for sourcing | 18:04 |
| smoser | well, it shoudlnt happen. can you get the full log ? | 18:05 |
| smoser | i'm nto sure why you're running twice. | 18:05 |
| smoser | the order of the log you pasted is really confusing | 18:05 |
| smoser | either the ln ran the second time *before* the 'running "init"' message | 18:06 |
| smoser | or the order is just flipflopped | 18:06 |
| smoser | (stdout/stderr possibly) | 18:06 |
| jlebon | smoser: journalctl output http://ur1.ca/ocj9o | 18:06 |
| jlebon | notice the bootcmd module running on line 1143 and on line 1427 | 18:07 |
| smoser | well notice line 996 and 1252 | 18:09 |
| smoser | init-local is running twice | 18:09 |
| jlebon | hmm, i thought the 1252 line was just to "close off" the logs for that instance | 18:11 |
| jlebon | they're the same PID | 18:11 |
| jlebon | the first bootcmd is run by pid 742 and the second by 856 | 18:12 |
| smoser | jlebon, so, what is suppose to happen is that cloud-init local should write a file that said that it ran . and the cloud-init-net shoudl basically do nothing | 18:17 |
| smoser | i really can'tdebug this much more right now, i'm sorry. | 18:18 |
| smoser | i do think you're right about the same pid though (i was wrong about local running twice) | 18:18 |
| jlebon | smoser: no worries, this is not high priority for me as well, just thought i'd investigate on here | 18:19 |
| jlebon | thanks for looking into it | 18:19 |
| jlebon | if it does become a bigger issue for me, i'll see if i can look into it | 18:20 |
| === shardy is now known as shardy_afk | ||
| === cpaelzer_afk is now known as cpaelzer | ||
| === cpaelzer is now known as cpaelzer_afk | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!