core20 snap stuck in auto-refresh

[Jeff Hillman]

20.04.3
snapd 2.55

Currently seeing a situation where the core20 snap got put into a disabled mode. running sudo snap changes shows that for 20 days it has been in an auto-refresh state.

running snap abort <id> puts it into an aborting state, but this has sat for over 24 hours now.

This is causing all snaps dependent upon core20 to fail. for example, running `lxc list` fails saying that there is no core20 file available.

More logs to come.

[Miguel Pires (miguelpires1)]

I've commented in the SF ticket but I'll summarise it here for clarity. The core20 snap was stuck in an auto-refresh because after copying data (as part of the refresh process), snapd called `sync` which just hanged for 25 days. Looking into the logs shows a lot of iSCSI "TCP connection closed" which is probably what's causing the fsync to never return.

[Alok Prudhvi (alokprudhvi)]

Hi Miguel, What can be the workload to get the core20 back to normal and get all the lxc list commands working?

Can you summarise the steps we could take to over come this problem?

[Miguel Pires (miguelpires1)]

Hi Alok. So far, the issue in snapd seems to come from a `sync` call to the file-system hanging indefinitely. If the root issue lies in the file-system, there's no workload that snapd can run to fix it. Given that rebooting made the problem go away and we can't test this further, there's currently no indication that snapd malfunctioned. So, for now, I'll mark this report as "Incomplete" so that we know that there's nothing we can do with the current information. However, if that changes in the future, we can pick up this report and go from there. Thanks.

[Alok Prudhvi (alokprudhvi)]

Thanks Miguel.

[Launchpad Janitor (janitor)]

[Expired for snapd because there has been no activity for 60 days.]