nickt
Contributor
- Joined
- Feb 27, 2015
- Messages
- 131
Hi all,
I had a power failure last night (yes, I know, I need a UPS - will be ordering one shortly). My FreeNAS server (latest stable 9.3 build) recovered OK. Pool and disks look fine (scrubs found no errors, SMART tests report no errors). Everything is working fine.
However, since the power outage, I get the same report every minute from autosnap.py in the logs, which suggests clean up of old snapshots is silently failing:
This happens for each one of my auto-snapshots (2 week retention), and the pair of log entries appears every minute. If I try to do the same at the command line, I get a silent fail from the zfs destroy command (i.e. nothing returned):
If I try to manually destroy the old snapshot in the GUI, I get a GUI message saying that the attempt failed as the dataset is busy.
Now, looking back at the logs, there is no question that this only started following the power failure, but I can't find any evidence of any kind of pool / disk corruption. Restarting my FreeNAS box hasn't helped.
Can anyone think of anything I could try? I assume the issue is that something is causing the snapshots to be marked as "busy", whatever that means. Anything I can do to "release" them? Is there possibly some kind of dataset "state" that may have become unsynchronised when the power failed?
Thanks in advance.
Nick
I had a power failure last night (yes, I know, I need a UPS - will be ordering one shortly). My FreeNAS server (latest stable 9.3 build) recovered OK. Pool and disks look fine (scrubs found no errors, SMART tests report no errors). Everything is working fine.
However, since the power outage, I get the same report every minute from autosnap.py in the logs, which suggests clean up of old snapshots is silently failing:
Code:
May 6 14:55:05 Saturn autosnap.py: [tools.autosnap:59] Popen()ing: /sbin/zfs get -H freenas:state "pond/Titan/Archive@auto-20150421.2300-2w" May 6 14:55:05 Saturn autosnap.py: [tools.autosnap:59] Popen()ing: /sbin/zfs destroy -r -d "pond/Titan/Archive@auto-20150421.2300-2w"
This happens for each one of my auto-snapshots (2 week retention), and the pair of log entries appears every minute. If I try to do the same at the command line, I get a silent fail from the zfs destroy command (i.e. nothing returned):
Code:
[root@Saturn] /var/log# /sbin/zfs get -H freenas:state "pond/Titan/Archive@auto-20150421.2300-2w" pond/Titan/Archive@auto-20150421.2300-2w freenas:state - - [root@Saturn] /var/log# /sbin/zfs destroy -r -d "pond/Titan/Archive@auto-20150421.2300-2w" [root@Saturn] /var/log# /sbin/zfs get -H freenas:state "pond/Titan/Archive@auto-20150421.2300-2w" pond/Titan/Archive@auto-20150421.2300-2w freenas:state - -
If I try to manually destroy the old snapshot in the GUI, I get a GUI message saying that the attempt failed as the dataset is busy.
Now, looking back at the logs, there is no question that this only started following the power failure, but I can't find any evidence of any kind of pool / disk corruption. Restarting my FreeNAS box hasn't helped.
Can anyone think of anything I could try? I assume the issue is that something is causing the snapshots to be marked as "busy", whatever that means. Anything I can do to "release" them? Is there possibly some kind of dataset "state" that may have become unsynchronised when the power failed?
Thanks in advance.
Nick