Looking for advice on settings for Backup from Storage Snapshot via iSCSI

[electricd7]

Hello-

I am working to enable backup from storage snapshots on my jobs which are running against a vSphere 7 environment which is backed by a Pure Storage array connected to each host via iSCSI. My Veeam proxies are virtual machines which I have enabled MS iSCSI and also installed MPIO on. I have configured 2 additional NICs on the Proxy servers and configured their IP addresses for the A/B side of the Storage Network and tested connectivity. I have installed the Pure Storage plugin and scanned the storage and verified that it connects via iSCSI. In this environment I have 135 virtual machines which are backed up daily in Veeam by one of 18 jobs (jobs are created one for each Folder in vCenter and VMs reside in one of 18 folders.) I altered a job and turned on "Enable Backup from storage snapshots" in the Advanced Settings for the job and also checked "Failover to standard backup" and "Failover to primary storage snapshot" on the settings but did not enable "Limit processed VM count per storage snapshot" checkbox. I am looking for a little clarification and advice as to how to tweak these settings for my 18 jobs. Basically I fire off the 18 jobs at various hours through the evening but I am certain that many of them run concurrently. Since all the VMs are housed within a single VMware datastore which coincides with a single Pure storage volume, snapping that volume essentially contains ALL 135 of my VMs. Can all 18 jobs share the same snapshot from another job which is still running if the job starts when another job is already running (ie no real need to snap again in my opinion if the storage snap was within the same hour or so)? Also what the "Failover to primary storage snapshot" setting do? What primary snapshot? And finally, I don't think MPIO is working properly as I see "Single controller connection" warnings for the proxies in the Pure panel when the jobs are running. What additional steps do I need to take so that the proxy makes use of MPIO and both storage network links while backing up? Thanks!

[david.domask]

Hi Chad,

Regarding the VM processing, first I'll link the relevant User Guide page -- the storage snapshots aren't shared between different jobs. So the limit is mostly about the concurrent VMware snapshots and the ESXi host load as per the User Guide link. How to configure this is dependent on the environment and also how your VMs are distributed across the datastores, but sharing a snapshot from one job to another is not currently possible.

The Failover to Primary Snapshot option refers to Backup from Secondary Storage Array; for storages that have replication relationships between the storage arrays, and Veeam will act as described in Step 5 there. The purpose of the option is to allow failover to reading from the Primary Array if the backup from Secondary Array fails for any reason, ensuring that _some_ backup is taken. This is optional as there are certainly situations when a storage administrator doesn't want the additional workload on the primary array.

The warning I suppose is about the proxy configuration when the temporary storage snapshot gets mounted; I would check this thread as it seems to be related to configuration of the proxies, but if it doesn't help, a Support Case will likely be best and let Support review the logs for one of the affected jobs. I don't think this is related to Veeam specifically so much as Veeam is triggering it, and the debug logs may give better clues as to what needs to be adjusted.

[electricd7]

Thanks. This all makes sense and I actually saw that post about MPIO and Pure but unfortunately the pure link leads to a dead page now. I will open a case with Veeam as there doesn't seem to be any real guidance on how to get Veeam to utilize multiple links when backing up from storage snaphots. This leads to issues such as veeam backup failures if we lose one of our storage array heads during a job or if we are performing an upgrade or something on the Pure during the backup window. It would be great if Veeam proxies would connect to the array in question over both accessible links and utilize MPIO during jobs to circumvent this. Anyway, I will open a case and ask support how to configure so this occurs.

[electricd7]

I was just informed by support (Case # 07619989) that Veeam does not support nor does it use MPIO when backing up from storage snapshot via iSCSI so no point in installing the feature for all who stumble upon this forum post. Thanks all.

[markdpolin]

Hi Chad,

You may want to open a support case with Pure too. I have heard from Pure that it is always necessary to provision a permanent dummy volume and connect it to the client. It doesn't need to be brought online, just connected to maintain the iSCSI sessions. plus, manually configure all the multipath sessions in the Windows iSCSI initiator so that when it's backing up from snapshot (volume copies) it will inherit all those sessions and thus not trigger those alerts.

[electricd7]

I see...so if I were to connect a dummy volume via MPIO inside of windows when Veeam creates and connects its initiators it may just connect using the same paths...thats a good idea. I will give it a shot and open a case with Pure Storage as well. Thanks!

[electricd7]

This was the trick. We had to create a dummy volume in the Pure array and then manually map each proxy server to that array via MPIO (video showing this process here https://www.youtube.com/watch?v=Bmf4cGZm0eo) which then showed each proxy in the pure with a connection to all 4 storage IP addresses. Once this was done, Veeam backup jobs were able to utilize multiple paths to storage from the proxy.

The Pure guide for setting up MPIO (use powershell portion) is here https://support.purestorage.com/bundle/ ... pplet.html.

Make sure to go into "Volumes and Devices" on iSCSI panel on Windows and select "Auto Configure" on the volume/mount so that it re-connects after reboot. it is not necessary to online/format the volume from within windows, it just needs the paths to work. This allows you to use a single iSCSI volume for all your proxies since iSCSI is a block level connector and you traditionally can't connect multiple windows machines to the same NTFS formatted disk via iSCSI.