HA Sync is very painful

[BillArchway]

Sorry if this is a duplicate. I see lots of talk about this topic but much of it is dated.

I'm using StarWind 6.0 with ESX. Very impressed. We lost one of our HA servers and esx never knew it happened.

My only issue is that the sync process is very painful. It takes hours. And you've lost HA redundancy for that time. And, sure, I inderstand that when you lose a node, it has to be synced to get the data to the new node. But it seems like almost anything you do to an HA node requires a sync. I'd love to chang my sync IP address scheme. But I certainly don't want to take a day resyncing. There isn't a way to edit the ip adress without deleting the partner and hence a sync?

I asuume I could stop all my VMs, delete and recreate the partner, manually mark both as synced and restart my VMs. But even that would be painful.

Can anyone explain what is the difference between a fast and full sync? A nice feature would be if HA realized when sync was broken and only synced updated blocks. Is that what fast sync is?

Thanks,

[anton (staff)]

Boosting sync priority puts more pressure on alive host (VMs running from it experience lack or delay of service). The best way to proceed is to use 3-way replication and not 2-way replication as losing 1/3 and not 1/2 from your storage back end cluster is much less painful.

P.S. BTW, V8 had improved a lot of things including sync priority mgmt.