=== jchittum_ is now known as jchittum | ||
AnhVoMSFT | rharper if you deploy VM, then add NFS mounts to /etc/fstab, then deallocate/start the VM again, mount -a will get executed | 13:42 |
---|---|---|
rharper | AnhVoMSFT: I guess the core issue is still that when cc_mounts runs networking *is* up; but on some OSes, the rpc daemon is not. In the case of a firstboot instance which adds an nfs mount, the mount -a -O no_netdev will fail to mount the requested mount; and since it wasn't already in fstab, systemd-fstab-generator won't create a mount unit for it. And clearly on your scenario the nfs mounts are present and systemd will | 15:48 |
rharper | mount them. However, if the rpcd issue were solved then mount -a would be perfectly fine thing to run exactly when it does. | 15:48 |
Odd_Bloke | falcojr: I've completed validation on https://bugs.launchpad.net/cloud-init/+bug/1922739 (and performed my share of the regression testing); where are you at? | 16:29 |
ubot5 | Ubuntu bug 1922739 in cloud-init (Ubuntu Hirsute) "AttributeError: 'DataSourceNoCloud' object has no attribute 'vendordata2_raw'" [Undecided,Confirmed] | 16:29 |
falcojr | Odd_Bloke: close...got a few failures that look unrelated, but re-running a few of them just to be sure | 16:29 |
Odd_Bloke | falcojr: OK, cool, once you're satisfied with your bug and regression testing (and have updated the tags on the bug), I think we're ready to ping an SRU person for release. | 16:35 |
falcojr | Odd_Bloke: Done! https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1899299 | 17:13 |
ubot5 | Ubuntu bug 1899299 in cloud-init (Ubuntu Hirsute) "do-release-upgrade from Trusty to Xenial breaks cloud-init init-local on reboot" [Undecided,Fix committed] | 17:13 |
=== bswinnerton0 is now known as bswinnerton | ||
falcojr | I pinged in #ubuntu-release | 18:16 |
aswinr | Hi. Looking to get this reviewed and checked in for the upcoming SRU https://github.com/canonical/cloud-init/pull/878 | 20:12 |
falcojr | aswinr: Thanks and noted. We'll try to get to it in the next couple days | 20:29 |
AnhVoMSFT | rharper: from rpcd perspective networking is not up because network-online.target has not been reached yet. Obviously there are distros dependent targets like systemd-networkd-wait-online that describes the point of "network is up, but just before network-online.target", but there's no guarantee there is such a target on every distro | 20:29 |
rharper | AnhVoMSFT: sure; even so networking *is* up; and if we skip nfs mounts with -O no_netdev ; there's nothing that will bring up nfs mounts on firstboot. | 21:35 |
rharper | the only other thought I had was to create a cloud-config.service.conf.d dropin which could use an ExecStartPre= script which could grep for _netdev in /etc/fstab, and if so, mount -a -O _netdev (bring up any nfs mounts if grep finds some) | 21:36 |
rharper | This would ensure nfs mounts are up prior to cloud-config.service running; | 21:37 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!