We noticed several volumes used for VM datastores were showing decreasing sis savings. When we checked there was a significant amount of space being consumed by the sis metadata. Netapp told us this is due to bug 657692 which is likely to be fixed in 8.1.2P4. The bug mentions "Following certain sequence of operations, this metadata can become stale".
The issue we have is the workaround is to run a "sis status -s" on each volume. When this is running it gradually removes the metadata which then causes the snapshots to increase rapidly. This means we can't automate the process as we need to kick it off and then increase the volume, snap reserve or remove snapshots.
Has anyone else experienced this issue or found a way to avoid the metdata becoming stale?
-- View this message in context: http://network-appliance-toasters.10978.n7.nabble.com/SIS-metadata-issue-tp2... Sent from the Network Appliance - Toasters mailing list archive at Nabble.com.
Yes - in this case we had enough space on the aggregate to allocate the space (sis status stuck at Fingerprint is being recovered from volume (0% complete))
http://www.gossamer-threads.com/lists/netapp/toasters/12815
Yesterday it happened on an aggregate without the space and it caused an outage (VMs crashed - spent the day cleaning it up)
I disabled dedup and without the bug fix I can not re-enable it.
On Apr 16, 2013, at 1:43 AM, Martin martin@leggatt.me.uk wrote:
We noticed several volumes used for VM datastores were showing decreasing sis savings. When we checked there was a significant amount of space being consumed by the sis metadata. Netapp told us this is due to bug 657692 which is likely to be fixed in 8.1.2P4. The bug mentions "Following certain sequence of operations, this metadata can become stale".
The issue we have is the workaround is to run a "sis status -s" on each volume. When this is running it gradually removes the metadata which then causes the snapshots to increase rapidly. This means we can't automate the process as we need to kick it off and then increase the volume, snap reserve or remove snapshots.
Has anyone else experienced this issue or found a way to avoid the metdata becoming stale?
-- View this message in context: http://network-appliance-toasters.10978.n7.nabble.com/SIS-metadata-issue-tp2... Sent from the Network Appliance - Toasters mailing list archive at Nabble.com. _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Short term, considering thin provisioning all the volumes, setting lower "almost full" thresholds and schedule "sis start -s" from a Windows host using powershell.
Longer term I think we'll have to go to 8.1.2P4 or 8.2 but would prefer a patch release than jumping to 8.2, last time we did this we hit another serious bug (SMB bug).
It seems like upgrading from 8.02P3 we've been on a endless cycle of Ontap upgrades to fix serious bugs.
On 16/04/2013 17:55, Fletcher Cocquyt wrote:
Yes - in this case we had enough space on the aggregate to allocate the space (sis status stuck at Fingerprint is being recovered from volume (0% complete))
http://www.gossamer-threads.com/lists/netapp/toasters/12815
Yesterday it happened on an aggregate without the space and it caused an outage (VMs crashed - spent the day cleaning it up)
I disabled dedup and without the bug fix I can not re-enable it.
On Apr 16, 2013, at 1:43 AM, Martin <martin@leggatt.me.uk mailto:martin@leggatt.me.uk> wrote:
We noticed several volumes used for VM datastores were showing decreasing sis savings. When we checked there was a significant amount of space being consumed by the sis metadata. Netapp told us this is due to bug 657692aa which is likely to be fixed in 8.1.2P4. The bug mentions "Following certain sequence of operations, this metadata can become stale".
The issue we have is the workaround is to run a "sis status -s" on each volume. When this is running it gradually removes the metadata which then causes the snapshots to increase rapidly. This means we can't automate the process as we need to kick it off and then increase the volume, snap reserve or remove snapshots.
Has anyone else experienced this issue or found a way to avoid the metdata becoming stale?
-- View this message in context: http://network-appliance-toasters.10978.n7.nabble.com/SIS-metadata-issue-tp2... Sent from the Network Appliance - Toasters mailing list archive at Nabble.com http://Nabble.com. _______________________________________________ Toasters mailing list Toasters@teaparty.net mailto:Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
-
Fletcher Cocquyt -
Martin -
Martin Leggatt