Are there any well-known performance issues with virtio nics with freebsd in a bhyve?

Whan I open man man and exit from it with Ctrl+C , there is no bash propmt echo on typing characters anymore and bash prompt does not go into new line anymore

I actually observed similar thing on linux too in bash, but it is that easy to reproduce in Smartos

try using the command [reset] and if that doesn't work use [printf "\033c"]

this might be helpful: superuser.com/questions/969044/how-…set-the-terminal-emulator-from-tmux

Similar thing happens on Linux, after entering nslookup and then exiting with Ctrl+C

virsh console or any serial console forwarding is notorious for that

it means, the app (man) is missing terminal restore code from SIGINT handler.

it's ssh session.

so the weirdest thing happened, couple of hours back i tried to import the debian-11 image, and it hung at the final stages whie extracting the image. i ended up ctrl-c'ing it, and trying to imgadm import again,but that failed because of the zfs filesystem left behind. so i tried to destry it, but it was busy so i destroy -f it, and it hung.. for about 5-10 mins, and then at some point i lost ssh connection. after awhile it came back -

machine had rebooted

ok, i try to import image again. this time it works. i'm busy w other stuff in the mean time, eventually i get to actually setting the vm up, and i connect to it over ssh. immediately i notice smt is off, the root prompt shows the uuid, nd not the hostname - like some process was aborted prematurely .. i try to run a few commands, most of them are hanging .. in the mean time i lose ssh access to the hst again - my urrent session froze,

and any new sessions hung

so i log into ipmi and see the console: mr_sas0: mrsas_tbolt_reset_ppc(): RESET FAILED, return failure !!! ... and mr_sas0: io_timeout_checker: FW Fault, calling reset adapter

i rebooted, same thing. reboot into recovery, run a scrub - no errors. write 20GB of random data using dd - no errors. i force-deleted the zone that was using the problematic image and imgadm delete the image, and ... things seem fine now

this makes no sense to me. even if the image was faulty or corrupted somehow, how can the errors be so severe to the point all i/o hangs

any suggestions?

Well, there could be a couple of things going on.

First, if you're really having hardware issues, that could explain all of this behavior.

Secondly, zfs recv doesn't do any of the standard zfs integrity checking. So if you've got a corrupted stream, it's going to assume that it's fine, but then bad things will happen later on.

It could also be a combination of these things.

is there some imgadm cache that could have been used at the second attempt? i wasn;t paying attention. but the import worked the 2nd time (after zfs destroy hang + reboot)

No, there's no cache.

i just imported the image again

no errors, but as soon as i create a vm based on it, bam, they begin

vmadm destroy hung too, it's back to recovery now to be able to untangle it. surely if it was smt with the image more people would have noticed (and even then i'd expect zfs errors, not driver errors?)

this seems like hardware problems, i just don;t understand why this image, reliably

Yeah, from what you're describing, hardware does seem like the culprit here.