When I updated to the latest kodi version (September?) on Monday 23rd of September 2024, Kodi ended up in a restart loop because the OOM killer killed kodi.bin
I needed a quick remedy and installed my last backup which, unfortunately, was one year agoā¦
That works quite well, but (at least) the youtube plugin doesnāt work anymore and anyhow, Iād like to have a newer version.
So what are my options?
- Is it possible to update to the August (or another version from this year) version of OSMC?
- Update to the current version and fight with the restart loop
- ???
Iād like 1. best, if that would be possible, since Iād have a working, almost current version. From there Iād create a backup and then maybe do 2.
Any suggestions?
Iād install an image with Kodi v20 and restore your backup; then upgrade to Kodi v21 via My OSMC.
Thatās a bit difficult: my backup is an image of the sd-card. Easy to install with dd, but it would overwrite the the Kodi v20 imageā¦
But if you took the backup of the whole SD card a year ago, it will be based on Kodi v20, so you should be able to upgrade from that, unless I didnāt understand you correctly.
Stupid question: I have a another raspberry with the same version that worked before the mentioned upgrade. So if I could use that one, my problem would be solved.
But: the messed up raspberry is a Pi4, the working one a Pi3 ā¦ I guess I canāt just use the image of the Pi3?
Yes, Itās a kodi 20.1
But if I upgrade via My OSMC now, Iāll get the version that caused the restart loop, wonāt I?
They are different images.
Well ā it depends what is causing the reboot loop.
If itās an OOM issue, you could try disabling some add-ons before upgrading; if that is causing your issue. Hopefully itās not being caused by the database upgrade process.
Ok, Iāll try that. After all, I still have last yearās backup
Thanks for your help!
Yes, it doesnāt hurt to try again.
If you are running additional services on the box you could also disable them temporarily to give Kodi more headroom.
I updated (via My OSMC) again, without disabling any addons, and went into the reboot loop again. So I stopped the mediacenter, added a 4GiB swapfile and started the mediacenter again. Memory consumption of kodi.bin went to ~3GiB (!) and the start took forever, but finally, it released most of the memory again (~0.7GiB).
Great, I thought, some conversion needed a lot of memory, but thatās done now - isnāt it?
After I restarted mediacenter again, it again used ~3GiB of memory
Next thing I tried: I moved the ~/.kodi/addons
to ~/.kodi/addons-suspicious
and created a new and empty ~/.kodi/addons
, hoping that some add-on is the culprit. Unfortunately, even without any add-on, memory consumption went to ~3GiB again
Iāve added a debug log of a startup here: https://the-grue.de/~markus/kodi.log.xz
I wasnāt able to detect anything iffy.
I canāt keep it like this: The startup takes ages and the swap file slowly breaks the SD-Cardā¦ I really hope you can help me.
Update: My video (and everything else) library is mounted via NFS. I unmounted it and restarted mediacenter: Tadaaa, normal memory consumption. Is something with the library update broken?
As soon as I mount it again and start a movie, memory consumption rises to ~3GiB again.
After all, it seems to me that Kodi 21 just needs much more memory on startup than Kodi 20. I donāt like this āsolutionā, but i bought a Pi4 with 8GiB memory and everything is āfineāā¦
BTW: i didnāt find any official memory requirements for OSMC. There is a page with supported devices, but that doesnāt mention the memory requirements. Did I look at the wrong place?
It should probably only need that for a large library migration; but not for actual usage.
It happens on every start of the Mediacenter and also sometimes during runtime. In the latter case, I think it was related to the movies library.
How could I check this?
My guess is the database migration is not yet fully completed.
I doubt itā¦
- Kodi consumes high memory on every start, then after some minutes consumption drops
- It has been started many times since
- it has been running for several hours each time
I would think that 1. means that migration is done, 2. would mean it had several attempts to do so and 3. means it had a lot of time to do the migration.
Is there a way check this? Like in the logs or in the database itself?
Logs should show if itās still doing stuff, yes.