Am 13.01.25 um 11:57 schrieb DERUMIER, Alexandre via pve-devel: >> Hmm, sounds like it might a bug, I can look into it. If really >> required >> to make it work, we can still set fixed node-names on the >> commandline, >> but also query them before usage to be sure we have the correct, i.e. >> currently inserted node. > >>> AFAICT, this is because the node name is not set in the original >>> options >>> for the block driver state and then it wrongly detects an attempt to >>> change the node name (even if specifying the correct auto-generated >>> one >>> during reopen). However, it is rather ugly to try and use a -drive >>> together with blockdev-reopen in any case, blockdev-reopen is really >>> written with -blockdev in mind and -blockdev necessarily requires >>> setting node-name up front. > > you are too fast ^_^ > >>> I don't think it's even worth fixing that bug. We should use >>> blockdev-reopen only after switching to -blockdev, it's much nicer >>> like >>> that 🙂 >>> >>> I'd still be in favor of querying the node-name of drives with >>> query-block before doing QMP operations though. Like that, we can >>> warn/error if it doesn't match what we expect for example, to catch >>> unexpected situations. > > ok, so the question is : how to query to node-noname of different disk > (+snapshot chain). > > Fabian is against the use of the path. > > So, I'm a little bit out of idea ^_^ >
For almost all QMP commands, we only need to care about the node that's inserted for the drive. And for your use-case, checking that the top node of the chain matches what we expect is already a good first step. The lookup itself is a different question, I'll also answer to the other mail. _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
