1GB channel.db cant be compacted on 32bit OS: mmap memory allocation error

[dunkfan]

Unfortunately i ran into this issue with no warning beforehand and now saw references to it in 2020/2021.
It would have been nice if lnd had notified regarding the issue but you still have a chance to help now :)

How is this best to be handled in 2025?
Upgrade the OS or would more swap space or something else handle it?

Running 0.18.5 on a 32 bit raspberrypi os:
Linux raspberrypi 6.1.21-v8+ #1642 SMP PREEMPT Mon Apr 3 17:24:16 BST 2023 aarch64 GNU/Linux
chantools was able to compact the file on a 64 bit system.
Compacted file now has 1118035968 bytes
TY

[Roasbeef]

Are you able to update to a 64-bit system? 32-bit systems have a fundamental memory limit.

Many years about raspiblitz migrated to a 64-bit system for this reason: #4811 (comment).

[dunkfan]

Thanks. I have a 64bit system but it only has 4GB ram.
Is that viable with enough swap on an SSD?

[hieblmi]

If that doesn't work you can

• shut down lnd
• back up the channel.db file
• compact it with chantools (which I saw worked from your PR description)
• start lnd with compacted channel.db file
• keep the compation on startup enabled so you don't run into this issue again.

[saubyk]

Closing this issue and moving the thread to a discussion if any further assistance is required.

[dunkfan]

I was not aware that lnd was auto compacting because of a setting in lnd.conf and that the node could come back online temporarily solving the startup problem simply by turning off the auto compact setting #db.bolt.auto-compact=1.
If lnd was interested in having lots of node runners (and lightning transactions) a little info here would have been useful.
Also the 32 bit problem could indeed be warned about on startup, before the next unsuspecting user falls into the trap(s)