Problems with automated Snapshots

[Patrick Neuner]

Hello,

we have configured some half-hourly snapshots für our Volumes - and also some remote Snapshots. In the last 2 days some of them were not executed properly.

They failed with this Error:

Subject: Snapshot schedule 'vol-vzct325_Sch_SS_2' status in cluster 'ClusterName' = 'Failed': 'The request was cancelled.'. WARNING (E00070100)

Event                   :              E00070100 EID_SCHED_STATUS_FAILED_MSG

Severity                :            Warning

Component               :       SAN/iQ

Object Type             :        Snapshot Schedule

Object Name             :      vol-vzct325_Sch_SS_2

Management Group  :    MgmtGrpName

Cluster                 :             ClusterName

IP/Hostname             :      node1Name

Date and Time           :     06/18/2013 22:25:03 GMT

User                    :               System

Resolution              :          http://h20000.www2.hp.com/bizsupport/site/search/r4_0/jsp/search.jsp?lang=en&cc=US&pc=0&sc=2&cf=1&cm=1&cs=1&c130=1&c115=1&c110=1&c120=1&nh=10&rf=0&lkc=1&lk=1&tx=E00070100

Message                 :          The snapshot schedule 'vol-vzct325_Sch_SS_2' status in cluster 'ClusterName' is 'Failed': 'The request was cancelled.'.

Was not able to find the cause of this Problem ... maybe someone here can point me in the right direction?

Thank you

Andreas Schnederle-Wagner

[Emilo]

Hello,

30 minutes is very, very aggressive you are probably running to many of these simultaneously.

I am not sure why you need to run them every 30 minutes?

You should look to back off a bit.

[Patrick Neuner]

Hi,

most of our Snapshots are done 6-hour and 1-day for remote snapshots.

The 30-Minute ones are only a few on some mission critical Volumes.

CPU/RAM usage is not high - can it really be that the snapshots are blocking each other?

[oikjn]

are these remote snapshots that are failing or local ones?  If remote then they might not have enough time to propogate before the next shap is initiated.

If these are application aware, then there might be a problem with the app-aware snapshot manager and that service might need to be restarted.

Can you give more details about the LUNs/apps experiancing the problem?  Given that your other schedules are working it seems like the direction of focus should first be on the initiator side.

[Patrick Neuner]

- only local Snapshots failed so far

- NON application aware

- completely random which Snapshot (Volume) fails

[Patrick Neuner]

btw) are there any "best practices" for "how many snapshots in which amount of time" are OK to configure?

Can't find any details in the official Docs ...

Andreas

[Emilo]

Yes 30 minutes is very aggresive.

Snapshots are a 'moment in time' picture of what the data was at that moment.

I understand mission critical but every 30 minutes?

I think you need to take a more realistic approach as to what is practical.

You can look at the 'mgrClientBinMon.log and you will probably see if yo do a search that is always deleting something.

[Patrick Neuner]

Hi,

I just downloaded the mgrClientBinMon.log and did not find any reference to deleted snapshots - could it be in another Log?

Also looked within CMC - deleting status is always very short - and not very oftern

Andreas