Connection broken on 1.13.1

[juditnovak]

Describe the bug Connection to multipass instance got interrupted.

Syndrome was observed after removing and re-installing the multipass snap. After rebooting the machine (host system, i.e. "laptop"), situation improved.

To Reproduce How, and what happened? I've been using a multipass instace with a fully functional Juju installation bootstrapped against LXD. (In case more info needed on the above, feel free to contact.) NOTE: the multipass instance was hosting a jammy OS.

1. multipass shell <instance>
2. cd <git_repo_local_checkout>
3. Running tests that include long delays on charm build.
4. The connection got closed.

The above happened multiple times. When running multipass shell after, on occasions the instance had to be started. Meaning that it has shut down (not only break connection).

Expected behavior Keeping up multipass shell fully functional even after long delays.

Logs Logs are difficult to follow as I've been destroying, re-creating, re-installing multipass. But I think this snippet may correspond?

12:48 pegasus systemd[1]: snap.multipass.multipassd.service: A process of this unit has been killed by the OOM killer.
12:49 pegasus multipassd[87076]: Received signal 15 (Terminated)
12:49 pegasus multipassd[87076]: Goodbye!
12:53 pegasus multipassd[87076]: Failed checking SSHFS mount "repos" in instance 'vm-lxd-22': ssh connection failed: 'No route to host'
12:56 pegasus multipassd[87076]: process state changed to NotRunning
12:56 pegasus multipassd[87076]: error: program: qemu-system-x86_64; error: Process crashed
12:57 pegasus dnsmasq[87151]: error binding DHCP socket to device mpqemubr0

Additional info

• OS: jammy, amd64
  • multipass version: 1.13.1 (both multipass and multipassd)
  • multipass info

Name:           vm-lxd-22
State:          Running
Snapshots:      0
IPv4:           10.38.193.57
                10.254.188.1
Release:        Ubuntu 22.04.4 LTS
Image hash:     1718f177dde4 (Ubuntu 22.04 LTS)
CPU(s):         12
Load:           0.08 0.25 0.27
Disk usage:     34.9GiB out of 145.3GiB
Memory usage:   3.9GiB out of 31.3GiB
Mounts:         /home/kaboca/repos => repos
                    UID map: 1001:default
                    GID map: 1001:default

• multipass get local.driver: qemu

[ricab]

As discussed elsewhere, you just went out of memory.

[georgeliao]

@ricab , is there a chance that it is related to the memory leak issue besides the actual consumption of the VM memory?

[ricab]

Hey @georgeliao: it is difficult to say for sure, but I think it is unlikely. There is no evidence of a leak in here and the issue you link was only reported on macOS so far. Plus, it happens over a period of time. In this case, it seems that the snap was updated and reinstalled over the course of a day. In any case, we already have that one open to investigate when we get a chance.