Third time I face the same issue since upgraded to Incus 7
I know it is strongly discouraged to use swtpm.
I have a home lab in which I have 2 IncusOS server (with a TPM1.2 chip which is disabled to use swtpm) running version 7.
Everytime one of them tries to upgrade, the installation goes just fine and then stuck in a reboot loop.
IncusOS relies on the TPM (physical, fTPM, or swtpm) to automatically unlock the encrypted LUKS volumes at boot. Once that happens and the system boots, the TPM isn’t used any further. Given that, I don’t think the use of swtpm should be affecting the running of Incus itself.
Is there any useful error message included when Incus fails to start? You could also try booting back into the prior version of IncusOS and seeing if everything starts up properly in that case. IncusOS 202605061755 was the first release to include Incus 7.0, which at least helps to eliminate some possible issues tied to Incus updates.
I have an old physical machine I run IncusOS with swtpm on, and just updated from the 4/28 to 5/12 release, with no issues that I observed. However, as you’re seeing this on multiple systems something’s going on….
PS – IncusOS releases are always versioned as YYYYMMDDhhmm; most IncusOS servers will run Incus, but not all of them, so it can be somewhat confusing to say things like “IncusOS 7”.
Thank you for the detailled response.
So since I know that it is not the expected behavior and that it is not supposed to affect any upgrade, I will continue digging in.
Are the two nodes clustered together or standalone?
Are you seeing it attempt to start Incus until the timeout (10 mins or something) and then fails and tries again?
Asking as getting stuck at that stage where the Incus daemon is trying to start can be caused by communication/quorum issues with other nodes in the cluster, at least that’s what I experienced recently.
You could also see if Incus gets to the point of serving on port 8443 at any time and attempt to grab debug logs:
