Hi all,
Currently looking for a quick crash/recovery method for an entire host on a Solaris 11 system, our procedure is as follow:
- Overall contents is backed up using TSM.
- OS image is backed up using UAR archive in recovery mode with --root-only flag.
AI is automatically configured to create regularly a profile/manifest from UAR archive (as a source image, and with root pool disks layout taken from this archive). So, in case of a complete recovery for a give host, we just have to boot net, and the operating system will be automatically restored (ZFS root pool, datasets (and datasets properties), and data content).
But for the other pools (applications, tools, other data, etc.), the procedure is to:
- Recreate manually the other ZFS pools.
- Recreate manually the other ZFS datasets.
- Restore from TSM for data content.
Those last steps are pretty tedious, error prone, and need lots of info (pools layout, ZFS lists and configured properties), all o that before being able to use TSM for data content restoration.
Q1: Is their an equivalent of the IBM AIX savevg/restvg commands which are able to save and restore layouts/properties taken from current pools/datasets (analogy to VG and LV in this case) ?
Q2: Is their a better way to get this stuff done (non-root pools and datasets) recreated automatically?
Note : Doing a complete UAR is not an option (because of size, coherency, etc. of the "user" (read non-root) data to backup/restore).
Thank you.
--
Best regards,
Julien Gabel.