StarWind Software

So, in my home lab, I'm in the middle of spinning up two vsphere 6.7 servers. Each host has a server 2016 VM as a storage appliance. I have a ton of ram on each server, so I thought to allocate 32GB or so to each VSA. Here's the thing: when we have a power failure, it's usually for an hour or longer, at which point the UPS has run out, and everything stops dead in its tracks. I had previously been using one vsphere host connecting to a CentOS 7 storage server with direct connect cables, with ZFS providing the storage. To improve write performance (using NFS), I had set sync=disabled, but that is problematic if an outage occurs. Since my lab has an APC UPS, I had apcupsd run a poweroff script when AC power fails, and a poweron script when it restores. The scripts enable or disable sync mode as appropriate. So, my question: is there any way to do this with Starwind? e.g. to have the VSAs query the UPS, and switch the L1 cache to write-through when AC power is lost, and switch back to write-back when it is restored?

Unfortunately, there is no way to dynamically change the operation mode of L1 cache in StarWind VSAN at present.

Ah, bummer. Thanks

You are welcome!

Hi,

while it's not possible to dynamically change the operation mode of the L1 cache what procedure could you recommend to perform this in a 2-node converged scenario?

Regards
bertram

Please find KB article by following this link.

Hi,

is it safe enough to disable the WB L1 cache on only one node to survive a following complete power failure or do you recommend to do it on both nodes?

In case of the second scenario it really becomes tricky because without a built in logic in Starwind it's hard to define which node starts with the process.

The amount of caching should be the same on both nodes.
The cache is HA as well.