Advice please: Restarted node with old lnddir state (accident)

[dunkfan]

I made the unfortunate mistake of restarting a node with the wrong lnd directory. It was a backup copy which was a few days old. Not many routing transactions had been made but some channel changes had taken place.
As expected that meant all channels were force closed and unfortunately I lost quite a few sats.
walletbalance is large but not where it should be :(
Currently my node is claiming all channels are gone (except for one private channel).

Notably, one of the new channels which was shut down does not show up in the "closedchannels" list.
Channel ID for that was: 942015383245488129 It was holding .1 BTC
Some advice would be welcome on what steps I should take - if any - to attempt to retrieve the lost funds.
For starters, should I restart the node with the correct lnd directory where for example the new channel existed? Could that help in retrieving the liquidity in that particular channel?

[guggero]

Did you force close any channels manually? If not, then you should be somewhat safe.
The missing channel is probably newer than the backup was, it was from less than a week ago.

I think it would make sense to start the node again from the new directory. Then check the lncli pendingchannels output. Things might only move when transactions are confirmed, so some patience will be required.
But all sats (minus fees) should come back eventually (given you didn't manually force close any channels while running on the old version).

[guggero]

You can turn off the compaction by setting db.bolt.auto-compact=false in your config.
But if compaction fails with that error, it means you have data corruption... So lnd will likely crash anyway, just later in the process... Seems like you're really being unlucky.

[dunkfan]

OK yes that failed quickly. I'm back up with the earlier version.
I have the channel.backup file with the correct channel state so perhaps using that would be the next unfortunate option?

[guggero]

Yes, using the channel.backup file should also allow you to sweep the closed channels (and ask the peer of the remaining open channel to also force close).

[dunkfan]

I used restorechanbackup with the incorrect but still working directory and channel.backup from the correct directory (Didnt use lncli create with my seed at all)
The .1 channel then closed leaving me a lot happier :)
There remains a private 'inactive' channel (not mine) where nothing moved for months with the liquidity on my side and 2 pending/waiting_close_channels. One of those has liquidity all on my side (zero transactions) and was known to both channel db files. Would using a force close here (the private inactive channel and the waiting ones) make any sense or should I just be patient?
Thanks so much

[guggero]

If you force close using an old state, that means you potentially could lose all funds if what you publish is an old state.
I would only do that if you are very certain the other party is never coming back (e.g. they shut down their node for good or also lost state). Unfortunately even then you can't be 100% sure they didn't use a watchtower.

So the safer route is to try to find out a way to contact the operators of those nodes and ask them to force close from their end.