On DHCI (2x Alletra 6050 + Witness on third site) Best practices when maintenance

[Daniel16]

Hello,

We need to break for some minutes the ISL dedicated for the iSCSI and Sync, is there some best practices or some attention to take care before the maintenance ? we know a partial resync will occur when link will be up, but except that is there any other attention ?

We are in DHCI Peer persistent, 2x Alletra 6050,  one array on each site/datacenter, with Witness on a third site, and ASO enabled.

DHCI is only used with Vmware, we haven't any VM in... i would say... "cross site state", i mean each VM residing for example on site "A", is computed on ESXi "A" and stored on Datastore/volume on "A".

the same thing for VMs on site (datacenter) "B" 

Thank you for your help

regards

Daniel

[Mahesh202]

Hi Daniel16

ISL is dedicated to the iSCSI and Sync meaning the replication subnet will be impacted. The following may occur

ASO of Sync volumes on the BGL to the GL expected

• The GL and BGL cannot communicate over the Group subnet.
• The Witness can still communicate with the GL and the BGL. The Witness's ability to see both arrays will prevent AFO from triggering.

Please refer to the below
https://infosight.hpe.com/InfoSight/media/cms/active/public/HPE_Nimble_Storage_Peer_Persistence_Deployment_Considerations.pdf

Hope this helps.!!

Regards
Mahesh

[SanjanaN]

Here are the steps customer can follow to perform ISL link maintenance:

Hi Daniel,

Here are the steps you can follow to perform ISL link maintenance:

Before ISL link brought down:

1. Disable ASO and unconfigure witness: to prevent unexpected switch over.

2. Edit all volume collections which have sync-repl volume configured: this will stop/pause volume replication activities 3. proceed with ISL maintenance 4. after ISL comes back up, edit volume collections to add volumes back for synchronous replication.

5. Enable ASO and configure the witness.

Hope i answered your question!

Regards,
SANJANA N

[Daniel16]

@SanjanaN Hi Sanjana,

Thank you for your explanation.

For the point 2 and 4, i don't understand, you mean to remove all replicated volumes on each volume collections ?

Do you think it's possible to simply "pause" de replicationpartners like in the printscreen below ?

Best regards

Daniel