Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Â
Guid snapshots with curly bracket found in NetApp Volume
1. Snapshots with curly {alphanumeric}
brackets found on the NetApp volume
Background: Recently I was investigating large number of âNetApp Volume Snapshotsâ on the 7-
mode FILER, which clearly looked like non-regular snapshots. I say non-regular b'cos normally FILER
does not allow you to create snaps with {} curly brackets, give it a try and it will throw an error.
Further into my investigation, I found out that there were hundreds of snaps across multiple
volumes hosting 'Exchange database' and along with {} snapshots there were also snaps such as:
'exchsnap__exch-srv1_MM-DD-YYYY_HH.MM.SS__daily' so that is an indication that there is an
application 'SnapManager for Exchange' behind the Exch DB backups. Hence, I found my source for
further investigation.
Source of the trouble:
Jumped to the server hosting SnapManager for Exchange application. Bydefault, SME works in
conjunction with VSS framework and SnapDrive, SME integrates with Microsoft VSS framework
seamlessly- i.e
VSS Components for SME Backup:
ï· VSS requestor = Backup Application âSnapManager for Exchangeâ
ï· VSS writer = MS Exchange writer âComes with your Application : MS Exchangeâ
ï· VSS Provider (SAN) = NetApp VSS Hardware provider <-API--> SnapDrive <-ZAPI--> FILER
Please note: Data Ontap VSS Hardware Provider component comes with SnapDrive. In other words,
when you install SnapDrive, VSS Hardware Provider also gets installed.
2. Typical SnapManager for Exchange Backup process from VSS framework perspective for Exch 2003
onwards: [Volume Shadow Copy Service (VSS) was introduced in Windows Server 2003]
The entire process is orchestrated by the VSS coordination service initiated by SME:
1. SnapManager determines which LUNs it wants to capture and verifies that Exchange
Server is present as a valid writer.
2. SnapManager initiates the shadow copy process.
3. VSS informs Exchange Server and the Data ONTAP VSS Hardware Provider that a shadow
copy is starting. Exchange Server stops writing to disk.
4. VSS ensures that NTFS is in a consistent state.
5. VSS requests the Data ONTAP VSS Hardware Provider to create a shadow copy.
6. The Data ONTAP VSS Hardware Provider requests SnapDrive to create a Snapshot copy of
the storage system volume that contains the specified LUN.
7. SnapDrive requests that the storage system create a Snapshot copy of the specified
volume.
8. When the shadow copy is complete, VSS returns NTFS to a normal state and informs
Exchange that it can resume disk writes.
So, what is Curly bracket snapshot: is basically 'Shadow copy ID', in other words a 'GUIDâ that
uniquely identifies a shadow copy' which is absolutely part & parcel of the SME Backup process.
Reason: When SnapManager for Exchange performs a backup, it creates the Snapshot copy using a
VSS-specific [GUID] name format, bâcos SnapManager for Exchange is integrated into VSS framework
and hence it is govern by VSS format, once the shadow-copy is created, 3rd
party application can ask
VSS to forget about the snapshot and do whatever they need to do, depending upon their backup
procedure.
Once the snapshot copy is taken successfully, it is then changed to its SnapManager-specific name
format. This happens automatically. One should be able to view the original VSS-specific name of the
Snapshot copy and the rename process in the SnapManager backup report which logs this name
change.
Examples of Snapshot copy names:
ï· VSS-specific Snapshot copy name:
{206f538c-ecd0-4799-90f9-baff5d8e0f15}
ï· SnapManager-specific Snapshot copy name:
exchsnap_exch-srv1_03-22-2006_17.53.33
3. Question: Why did these GUID snaps were not renamed in this case and were left
âorphanedâ?
Answer: Looks like SnapDrive at some point was unable to communicate with Storage
system (Initiator lost connection to the specified LUN or Snapshot is inconsistent or does not
exist for some reason) and if any of these instances occur, the Snapshot will not be deleted.
I traced this error in the Windows Application event log: In-band SCSI command from LUN (FC) to
storage system returned invalid data. Snapshot doesn't exist.
As it turns out there is a known BUG: 808158
https://mysupport.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=808158
BUG says: Sometimes when deleting backups after creating backups especially in identification and
cleaning up of orphaned Snapshots of either a database or log volume, Snap Drive might not delete
the specified Snapshot because of any of following reasons,
ï· Snapshot is inconsistent or does not exist for some reason
ï· Initiator lost connection to the specified LUN
ï· A Data ONTAP issue
If any of these instances occur, the Snapshot will not be deleted yet Snap Manager for Exchange
deletes its SnapInfo folder (for example, meta-data of the Snapshot from the log volume) as the first
step toward cleaning up the orphaned Snapshot. As a result, the orphaned Snapshot might be still
not cleaned up properly.
Workaround: There is no permanent solution but only workaround - Manually go to the specified
volume (where that DB LUN exists) and delete the orphaned curly {Snapshot}, which was not
cleaned up properly because SnapDrive was unable to delete it.
ashwinwriter@gmail.com
July, 2018