I would like to capture the alternative discussion here as well. I've requested some more debug info in the bug linked.
IIUC, in the scenario described, the ephemeral disk may be replaced with one that is no longer formatted with ext4, so the existing entry in fstab for the ephemeral drive will no longer mount. As soon as rootfs is mounted, the systemd generator for fstab runs and will create .mount units for each entry. I want to capture those errors to confirm that's what is really going wrong. To the solution, I had suggested that in cc_mounts.py that the logic which is ensuring we have an entry in /etc/fstab for the supplied disks, could also check if the device is mounted, and if not, enable the 'mount -a' command in the activate_cmds list. While we're only seeing the issue on Azure; it's possible that this would affect other platforms with a similar issue and fixing it in a common path would be preferable to adding additional per-datasource config. -- https://code.launchpad.net/~jasonzio/cloud-init/+git/cloud-init/+merge/366329 Your team cloud-init commiters is requested to review the proposed merge of ~jasonzio/cloud-init:requestMountAll into cloud-init:master. _______________________________________________ Mailing list: https://launchpad.net/~cloud-init-dev Post to : [email protected] Unsubscribe : https://launchpad.net/~cloud-init-dev More help : https://help.launchpad.net/ListHelp
