Do snapshots with bhyve actually work? I enabled the ability on the headnode, and I can issue the "triton snapshot create" command on a new bhyve instance with "Flexible Disk" selected. But the job ultimately fails with: vmadm.create_snapshot error: vmadm exited with code: 1 signal: null

I'd love to use LX containers more, but I just run into so many weird problems with them.

@jdt yes, we have it working at mnx. I'll have to dig up further details tomorrow, but if I recall, you'll need to have packages sized with a quota at around 2x the disk size to allow for snapshot space.

jdt: That should be fixed in release-20231116 and later.

jdt: Oh, also the guest needs to fstrim regularly.

Okay, thanks. I did just update all of the zones and platform images to the latest this week. bahamat like a cron job to fstrim? And mwilkens, let me know what you find tomorrow - thanks!

jdt: Yeah. All of our current HVM images have fstrim cron'd by default.

Okay, cool.

I just installed smartos, but wheng I do imgadm avail it hangs forever.

I just saw an old bug reappear, 'ttymon: set_termio_ get_ttymode failed: Timer expired' / 'ttymon: initial termio on (/dev/wscons) failed', when booting up a server to which I downgraded to 20231130T043122Z following the bug on the Mellanox CX4 cards.

Any suggestions?

xmerlin: Off the top of my head, I say wait for tonight's release spin, only so future debugging can be up to date. (Yes the CX-4 fix will be in there.)

danmcd, thank you very much

speaking of that, should I ping Alex to see if he wants to take a look at it?

Yeah, though he might have new-semester overhread.

i'll shoot him an email since he's not on irc recently

I did get a bhyve snapshot to work. But I'm a bit confused about bhyve flexible disks. I adjusted the quota to 20GB (it was at 10GB) and kept a flexible disk assigned at 10GB. The image disk seems to be 10GB as well. So I suspect that it's just the act of making the quota twice as big as the inherent instance drive that allows for snapshots - is that correct?

Secondarily, the additional disk shows up as assigned in adminui, but is nowhere to be seen in the instance itself (i.e., there is no /dev/vdb and 'triton instance disk list' shows only the one 10GB drive). I'm not sure what to make of that.

This is a debian-12 (20231127) instance fwiw.

Hmm. Or maybe what I'm seeing in adminui is just the primary disk. Looking at the PCI_SLOT, that is what shows up in lspci. So I'm probably just misunderstanding the configuration. I thought "adding" a disk to the bhyve package would result in an additional disk, but it looks like it's just sizing the primary disk. I imagine adding a second one would show up as a separate drive.

Yep, that seems to be it. "Quota" represents the space required for all disks. "Disks" represents all disks on the system, inclusive of the primary, system disk (that's where I erred - I thought the system disk was separate). "Quota" needs to be about twice as large as the space allocated for all of the disks defined in "Disks".