=== thnee1 is now known as thnee [15:14] https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1671951 [15:14] Ubuntu bug 1671951 in systemd (Ubuntu) "networkd should allow configuring IPV6 MTU" [Medium,Confirmed] [16:54] rharper: per your review coments on my mtu branch (in _extract_addresses) what's our desired behavior if config represents on mtu key at the physical device level and one in an ipv4 subnet? [16:54] as it is currently, the subnet mtu would be overwritten by the top-level mtu [16:59] I could change your review comment to only honor device-level mtu setting if an mtu(ipv4-specific subnet setting) doesn't already exist === Rolf is now known as Guest91333 [17:03] Hi, I am trying to write a part-handler, version2, but even a simple import os will make the part-handler stop working. No printout in the logfiles from code after the import os. Any pointers what might be wrong ? [17:05] I am using cloud-init 0.7.5 [17:06] Guest91333: 0.7.5 is very old. i'd suggest first seeing if you can get something working with more recent. [17:07] either way though, a fix to 0.7.5 is not really l ikely [17:07] blackboxsw: hrm, not sure; there can be only one v4 mtu (aka, device level) it's not clear to me which one we should respect; I think I'd like to log that we found conflicting values and pick one [17:07] upgrading the cloud-init version is difficult, as the image is given to us. Anyone has an example except the one in the docs ? [17:09] if on redhat/centos : https://copr.fedorainfracloud.org/coprs/g/cloud-init/el-testing/builds/ has our builds [17:12] hrm Guest91333, if you are adding a part-handler to an image you don't control content on, I'm having a hard time seeing how that part-handler will be executed as cloud-init will only run on initial clean boot in that environment. [17:13] unless you are running a script after the fact I guess [17:14] mime type part-handler - oh, you call more than one thing a part-handler ? Sorry forthe confusion then ! [17:16] my code is based on this https://github.com/cloud-init/cloud-init/blob/master/doc/examples/part-handler-v2.txt [17:22] ohh gotcha, ok so you are generating a text/part-handler mime part. ok. now I'm with you, yet I haven't walked through this before. And yes, there have been significant changes since 0.7.5 that may have impacted the mime handling. :/ [17:23] If there was some issue importing modules in your mime part-handler part I'd expect to see some error log along the lines of "Failed at registering python file:" with the mime filename you supplied. Also there could be stderr over in /var/log/cloud-init-output.log that could help [17:25] "Failed at registerting python file .* \(part handler\).* looks like. Anyway that's about as deep as I go on that issue at the moment [17:25] :/ [17:25] I didn't see any errors in /var/log/cloud-init-output.log - just missing output from printouts after the e.g. import os. Also other logs in /var/log didn't show anything suspicious. Checking if we can update our image now. [19:16] powersj: ci fixes for pylxd ... needed serially installed before installing pylxd as latest urllib3 that came out 2 days ago results is a contextualdependency conflict when installing pylxd's sub-deps [19:16] https://code.launchpad.net/~chad.smith/cloud-init/+git/cloud-init/+merge/347637 [19:17] this broke all ci runs on cloud-init with the following: https://jenkins.ubuntu.com/server/job/cloud-init-ci/53/console [19:17] * blackboxsw shoots to get rid of at least one red status ball on jenkins today [19:18] ugh [19:28] powersj: I just reviewed https://github.com/lxc/pylxd/pull/309#pullrequestreview-126924422 I'm not sure that tip of pylxd has the fix right. though I'm not positive what the fix should be [20:19] thx for the review powersj , it should fix a couple of jenkins jobs (like cloud-init-integration-nocloud-kvm-x) too [20:19] just kicked it off to confirm [20:22] blackboxsw: thanks! yeah we have a few more red dots to take care ;) === Rolf is now known as Guest91018 [20:45] got disconnected - sorry if this is posted again. Short update from the mime-type part-handler. Seems it was a problem with the logging. The example uses print, which seems to be swallowed after so many bytes. Switching to loggers helped. Would be nice if the example in https://github.com/cloud-init/cloud-init/blob/master/doc/examples/part-handler-v2.txt could be updated to use loggers. Thanks ! [21:06] Guest91018: well merge proposals are welcome for doc for sure === r is now known as Guest32630 [22:07] another update from mime-type part-handler: it wasn't the logging, it was me testing with os.getlogin() which threw an exception. print works fine. Sorry for the noise. After reading explanation on stackoverflow, all makes sense now