Hi Michael,

I am not using ZFS and have plenty of free diskspace:
Filesystem     Size    Used   Avail Capacity Mounted on
/dev/ada0p2    190G     92G     83G    52%    /

top looks like this on a not-hung system:

last pid: 38292;  load averages:  0.91, 0.87,  0.88    up 0+01:13:16  13:12:12
32 processes:  2 running, 29 sleeping, 1 zombie
CPU:  5.5% user,  0.0% nice,  7.8% system,  7.0% interrupt, 79.7% idle
Mem: 59M Active, 3549M Inact, 1090M Wired, 610M Buf, 2250M Free
Swap: 24G Total, 24G Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE    TIME    WCPU COMMAND
38079 root          1  22    0    99M    43M RUN      0:03  14.02% pkg-static
38285 xi            1  20    0    14M  3612K RUN      0:00   0.10% top
37889 root          1  22    0    15M  3680K nanslp   0:02   0.05% sh
  759 root          1  20    0    17M  6200K select   0:04   0.01% tmux
37881 xi            1  20    0    21M  9252K select   0:00   0.00% sshd
  600 ntpd          1  20    0    17M  4980K select   0:00   0.00% ntpd
  749 xi            1  20    0    21M  8804K select   0:03   0.00% sshd
37882 xi            1  20    0    15M  5584K pause    0:00   0.00% zsh
37603 root          1  20    0    15M  3900K select   0:02   0.00% sh
78529 root          1  26    0    15M  3892K piperd   0:00   0.00% sh
36203 root          1  47    0    23M    10M wait     0:00   0.00% perl
  750 xi            1  25    0    15M  5296K pause    0:00   0.00% zsh
  542 root          1  20    0    13M  2632K select   0:00   0.00% syslogd
  341 root          1  20    0    11M  1424K select   0:00   0.00% devd
  753 root          1  20    0    16M  4428K pause    0:00   0.00% csh
37826 root          1  20    0    21M  9228K select   0:00   0.00% sshd
37790 root          1  52    0    15M  3896K wait     0:00   0.00% sh

It occurred to me that I could try using ahci-hd instead of virtio-blk, so I am trying that now to see if it makes a difference.

Best regards
Christian


On 5/3/21 12:31 PM, Michael Gmelin wrote:


On Mon, 3 May 2021 11:14:45 +0200
Christian Stærk <[email protected]> wrote:

Hello,

I recently upgraded one of my bhyve guests to FreeBSD 13.0-RELEASE.
The host is running 12.2-RELEASE-p4.

The guest in question is used for building packages with poudriere
and is always the first one to get upgraded to new releases.

The guest freezes consistently every time poudriere tries to build
the thunderbird package. Although it is not a complete freeze. It
responds to ping and programs like top and systat continue updating.
However, all I/O operations seem to hang.

systat looks like this when it happens:
https://borderworlds.dk/~xi/FreeBSD%2013%20frozen.png
<https://borderworlds.dk/~xi/FreeBSD%2013%20frozen.png>

What is the best way to debug this?
Sharing top output might help (memory/swap and ZFS arc usage). Also,
make sure you're not low on diskspace (these are the two most common
issues I encountered with hanging bhyve VMs - out of non-virtual memory
and out of diskspace of the host machine filesystem).

Some basic info on the host and your VM configuration might also help
others to reproduce the problem.

-m


_______________________________________________
[email protected] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-virtualization
To unsubscribe, send any mail to 
"[email protected]"

Reply via email to