3 ms·

I come from the GPFS, Lustre, and Panasas world of HPCC. Personally, I used (past tense) ZoL in 2014-2017 on Ubuntu. The issue is that the array eventually ent

I come from the GPFS, Lustre, and Panasas world of HPCC. Personally, I used (past tense) ZoL in 2014-2017 on Ubuntu. The issue is that the array eventually entered an unrecoverable state where it could no longer be mounted RW. That wasn't the end of the world, but the support from ZoL was to shrug at it. That was the end of that because without support and without pride, something that appears shiny is effectively useless. Been running many XFS volumes over mdadm RAID10 near-2 arrays. Zero major problems in 5+ years with over 400 TiB online. SGI's, Redhat's, and more contributions to various Linux storage components are excellent. My conclusion is that ZoL != Solaris ZFS. Once it left enterprise with controlled hardware and dedicated engineering & support teams, it regressed and devolved. Beware of fanboys where passion and tribalism exceeds evidence and reliability.

db48x 3y ago
If you had a Solaris support contract then they might have had an engineer poke around in your dead filesystem with `zdb` to see if they could salvage anything. `zdb` still exists, so you could have done that yourself. Why didn’t you?
rincebrain 3y ago
I'd be curious to know what you mean by "shrugged at it", with a link to the relevant bug. I would suggest that you should probably have a support contract with _somebody_ if you want a guaranteed turnaround time on responses. Plenty of people are paid to work on OpenZFS, but if you're not the one paying them, then it's always going to be a best effort thing. Since the incorrect code dates back to Sun, I don't know that you should be using this specific bug to cast stones.