Trash empty after e2fsck

So after a reboot the OS suggested to perform fsck.
I performed e2fsck and confirmed the suggested fixes.
Check finished.

Now node won’t start. Structure in trash needs cleaning. When I check the trash, it is completely empty. Even the satellite folders have gone.

I found a e2fsck folder on another partition. It contains files like a9fc6752-173c-4e6f-8062-51932b8530b7-dirinfo-M8ia7X and a9fc6752-173c-4e6f-8062-51932b8530b7-icount-dDWFm4.
I don’t know what these files are, never seen them before. And I don’t know what if they are of any use. How can I get the node to start again? I wonder if a second e2fck could help or worsen the situation.

So it looks like the files that have been created are only “scratch” files and not like some undo files to revert the changes.
Weirs, basically what I wanted, all trash no gone on this node but a different way, together with all satellite folders. Looks bad.

So next step is to run another e2fsck maybe that brings something back.

this is mean that you need to run fsck several times, until it would fix all errors and would report that everything is clean.

Thanks. I am doing that right now.
Unfortunately it is very slow.

1 Like

I know, but unfortunately there is no other solution so far. It requires several runs (as chkdsk in Windows by the way) to clean all problems.

The process interrupted due to low memory.

I am using already scratch files and I made a swap of 4G.

Any ideas how to make this complete?

Seems connect it to the more powerful Linux PC. Or keep trying.

By the way, if that’s PI, I think you need to enable using swap.

It’s an ODROID HC2 which is almost the perfect thing for a node. But it is a bit outdated now. They would need to build a new version of it. It has very limited RAM.

Yes, I do like this tiny thing a lot. Unfortunately they seems stopped to manufacture them and do not have a similar device but with more RAM as far as I know.

They would need modern version of it, 64bit, MMC instead of SD, min 4GB RAM, faster CPU.
The device itself is ingenous.

1 Like

In any case, please check the documentation, perhaps you need to enable swap explicitly somewhere and maybe extend it (it would be easy, if you would have LVM there).

That’s my swap on it currently running:

/dev/zram0 partition 995.2M 967.1M    5
/swapfile  file          4G 629.7M   -2

And I have scratch files enabled.

I see. With that setting I wouldn’t expect the OOM. However, if the disk is dying, then, well, you need much more and preferable on a different disk. Perhaps you need a greater SD card to place swap there?

By the way, my Pi3 B+ node has had 32GB SD card, so I did use it for swap (I know, and as a result it’s down now and I’m unable to go to that place to replace or re-flash it…).

Maybe that’s an idea. My SD card is 64GB 60Gb free. Maybe I should put the swapfile on there with like 16GB?

Yes, I would try it. Later you can move the swap out of the SD card to extend its life.

As it is currently running I don’t want to interrupt it.
But if it fails again or needs additional runs, I will change that and move the swapfile onto the SD card.

1 Like

Oh no, crashed again due to memory.
I have now the swapfile on the SD card and currently running it again.

These are exactly the things why I consider the ODROID HC-2 are outdated and would need some modern parts.

Yep, I can understand. Sorry about problems your node is currently suffer.

try odroid HC4 has 4gb ram and 64 bit.

It’s not the same. It looks more like a cradle

than as a this nice thing:

1 Like