[Bug 241083] zfs: zpool import seems to probe all snapshots
bugzilla-noreply at freebsd.org
bugzilla-noreply at freebsd.org
Sat Oct 5 14:13:34 UTC 2019
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=241083
Bug ID: 241083
Summary: zfs: zpool import seems to probe all snapshots
Product: Base System
Version: 12.0-STABLE
Hardware: Any
OS: Any
Status: New
Severity: Affects Only Me
Priority: ---
Component: kern
Assignee: bugs at FreeBSD.org
Reporter: d8zNeCFG at aon.at
Scenario:
- Homegrown server, 6 disks
- Each disk partitioned: ca. 3/4 for the first zpool ("hal.1"), ca. 1/5 for the
second zpool, rest for misc. stuff
- The first zpool is imported at boot (from unencrypted "3/4" partitions), it
contains about 300 zfs filesystems + 20 volumes, where the latter are
frequently partitioned (MBR, BSD, MBR + BSD, because they are used as iSCSI
disks for virtualized installations of various operating systems)
- The second zpool is manually imported (when needed) from the encrypted
partitions (each of the "1/5" partitions is encrypted), it contains about 20
filesystems (incl. volumes)
- Backups are kept by regularly taking recursive snapshots on both pools, about
10 backups are kept at any time
- This results in over 500 device entries in /dev/zvol (e.g.,
"/dev/zvol/hal.1/1/vdisks/925 at backup.2019-09-21.21:59:27s2a")
- At some point, the second zpool is imported
Result:
- The import takes a long time
- Using "gstat -a", it can be observed that there is much activity on each of
the devices in /dev/zvol/... during the time it takes to import the second pool
Expected result:
- Importing a pool should not look at snapshot but only "normal" devices, where
"normal" could be defined either as not containing a "@" or as being read/write
or by trying the devices with the shortest path first (or maybe there is a
better discriminator) in order to speed up the import
- Maybe (by default) it should not look at devices in /dev/zvol at all because
normally one would not want to create a pool from devices in another pool
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the freebsd-bugs
mailing list