Oracle ZFS Storage Appliance: Unable to Remove a ZFS Zombie Snapshot

Asset ID:	1-72-2021341.1
Update Date:	2018-01-05
Keywords:

Solution Type Problem Resolution Sure

Solution 2021341.1 : Oracle ZFS Storage Appliance: Unable to Remove a ZFS Zombie Snapshot

Related Items


Sun ZFS Storage 7420
 Sun Storage 7110 Unified Storage System
 Oracle ZFS Storage ZS3-2
 Sun Storage 7210 Unified Storage System
 Oracle ZFS Storage ZS4-4
 Sun Storage 7410 Unified Storage System
 Sun ZFS Storage 7120
 Oracle ZFS Storage ZS3-4
 Sun Storage 7310 Unified Storage System
 Sun ZFS Storage 7320
 Solaris x64/x86 Operating System

Related Categories


PLA-Support>Sun Systems>DISK>ZFS Storage>SN-DK: 7xxx NAS

In this Document

Symptoms

Changes

Cause

Solution

References

Created from <SR 3-7692652355>

Applies to:

Sun ZFS Storage 7420 - Version All Versions and later
Sun ZFS Storage 7120 - Version All Versions and later
Sun ZFS Storage 7320 - Version All Versions and later
Oracle ZFS Storage ZS3-2 - Version All Versions and later
Oracle ZFS Storage ZS3-4 - Version All Versions and later
7000 Appliance OS (Fishworks)

Symptoms

Existence of zombie snapshot which can not be removed.

We are still seeing zombie snapshot despite running firmware code with a fix to Bug 15783537

Manual ZFS destroy on Solaris shell failed with dataset is busy

On test machine:

nas-7110-1# zfs destroy -f pool-0/local/default/fs@.zombie-10ba
cannot destroy 'pool-0/local/default/fs@.zombie-10ba': dataset is busy

Changes

.zfs directory is set to visible

Cause

Race condition caused by another thread accessing a file or directory in the snapshot and at the same time scheduled snapshot is removing oldest snapshot.

ZFS project has property snapdir set to visible.

In this customer's case, an NDMP thread (could be any thread) was accessing .zfs directory when scheduled snapshot was removing oldest snapshot.

From supportbundle cifs/cifs.out:

pool02a/local/eb_gah12_d_04a/applmgr snapdir visible

Solution

Engineering is working on code fix.

As a workaround - reboot the appliance.

Other workarounds:

- Unmount the corresponding zombie dataset using Solaris shell (follow the steps in Bug:17330892)
- Re-import the related zombie pool

References

<BUG:17330892> - ZOMBIFYING SNAPSHOT RENAME FAILING DUE TO REMOUNT IN THE MIDDLE OF THE RENAME
<BUG:15783537> - SUNBT7159038-SOLARIS_11U1 UNABLE TO DELETE ZOMBIE SNAPSHOT CAUSES CPU SPIKES

Attachments

This solution has no attachment