Aura H2O Can I stop a Kobo H2O from marking the file system as dirty short of a factory reset?

[ceridwen]

I have a Kobo Aura H2O with a file system that always shows up with the dirty bit set, so when I plug into a Linux computer, it's mounted as read-only, which makes it impossible to add books. I've run fsck on the file system and didn't find any errors, so I'm unconvinced there's anything actually wrong with it. I can manually unset the dirty bit and remount the file system read-write and successfully add books, but the moment I unplug it and plug it back in, the file system is set as dirty again and I have to go through the whole remount procedure to make it usable. I'm assuming the Kobo software is responsible for setting the dirty bit. Is there any to force it to stop doing that short of a factory reset?

This problem seems to have started when my ereader ran out of charge while I was temporarily unable to use it.

[NiLuJe]

Err, fix whatever mess is mounting stuff ro if the dirty bit is set, because that's definitely not standard.

EDIT: I would have naively assumed that'd be because of errors=remount-ro, but that's the default on FAT, and it's not tripped by a dirty bit on my system.

----

Is that an actual H2O, first of its name, or a v2? Because that'd be the first time I've seen this being reported on a H2O (it's usually far more prevalent on devices with an eMMc).
I've certainly never seen it with mine (whereas it's fairly common on the Forma. Possibly slightly less so in newer FWs).

[JSWolf]

Try plugging the H2O into a Windows computer and if it thinks there is a problem, run check disk on it.

Always properly eject the H2O from the computer. You may have unplugged it without ejecting to get this issue.

[ceridwen]

Originally Posted by NiLuJe

Err, fix whatever mess is mounting stuff ro if the dirty bit is set, because that's definitely not standard.

EDIT: I would have naively assumed that'd be because of errors=remount-ro, but that's the default on FAT, and it's not tripped by a dirty bit on my system.

----

Is that an actual H2O, first of its name, or a v2? Because that'd be the first time I've seen this being reported on a H2O (it's usually far more prevalent on devices with an eMMc).
I've certainly never seen with mine (whereas it's fairly common on the Forma. Possibly slightly less so in newer FWs).

errors=remount-ro is what's triggering the load as read-only, I've assumed. That's set in mtab, IIRC. You mean the dirty bit isn't sufficient to trigger errors=remount-ro? Is there a way to figure out what is triggering it to go read-only, then?

This is an original H2O.

Try plugging the H2O into a Windows computer and if it thinks there is a problem, run check disk on it.

Always properly eject the H2O from the computer. You may have unplugged it without ejecting to get this issue.

I don't have a Windows computer. fsck is the Linux equivalent to checkdisk, AFAIK.

I always do properly eject. In this case, the issue definitely started after I didn't recharge it awhile and it discharged completely. (I wasn't using it during this time.) I'm guessing that effectively caused a crash and that's why it's marking the disk as dirty.

[JSWolf]

So run fsck and fix the problem.

[davidfor]

At the moment, I'm noticing an error like this if after connecting either my Forma or Clara HD to my work Linux machine. When I get home, Windows will complain that the file system is dirty. The check doesn't find anything. I assume the Linux box isn't unmounting something properly, but, it isn't enough of an issue to chase down.

If it is happening all the time, it could be failing storage. Others have reported this and replacing the card fixed it. That is only possible if it is an original Aura H2O. But, you will need to break the waterproofing to do that.

[PeterT]

For a long time I've experienced a dirty bit being left set when I dismount from a Chromebook. It doesn't cause an issue on the Chromebook but the next time I connect to windows the message is presented and requires a chkdsk to clear.

[trampas]

I had this every time with my Aura H2O 2nd edition even after running the windows check/fix. Eventually I tried the method in the link and no more problems!
https://www.raymond.cc/blog/manually...hout-chkdsk/2/

[JSWolf]

Sounds to me like a bug in Linux. This is yet another reason to use Windows.

[frostschutz]

the kobo uses "lazy umount" before USB export. unfortunately the way lazy umount works is that the filesystem "disappears" but in reality nothing is umounted at all if the filesystem is busy. so kobo ends up exporting the filesystem in dirty state.

in original unpatched kobo firmware this used to be the case when connecting it to USB out of a kepub book. but I haven't tested this condition recently so I don't know if this is still the case. if you modded the device it's possible for those mods/addons to block umount as well.

I tried to make a debug utlity for this condition at some point https://www.mobileread.com/forums/sh...d.php?t=263674 but it probably won't work anymore in current firmware, haven't updated in ages

[Quoth]

Wait till the Kobo ereader has removed the progress bar for updating content before unplugging USB.
I've had no problems in the last two years using Linux Mint with 3 models of Kobo, Nook, 5 models Kindle and two Sony models.
I've occasionally plugged into a Win7 PC last year.

[JSWolf]

Originally Posted by frostschutz

the kobo uses "lazy umount" before USB export. unfortunately the way lazy umount works is that the filesystem "disappears" but in reality nothing is umounted at all if the filesystem is busy. so kobo ends up exporting the filesystem in dirty state.

in original unpatched kobo firmware this used to be the case when connecting it to USB out of a kepub book. but I haven't tested this condition recently so I don't know if this is still the case. if you modded the device it's possible for those mods/addons to block umount as well.

I tried to make a debug utlity for this condition at some point https://www.mobileread.com/forums/sh...d.php?t=263674 but it probably won't work anymore in current firmware, haven't updated in ages

Windows handles this with no problems at all. +1 Windows.

[davidfor]

Originally Posted by Quoth

Wait till the Kobo ereader has removed the progress bar for updating content before unplugging USB.
I've had no problems in the last two years using Linux Mint with 3 models of Kobo, Nook, 5 models Kindle and two Sony models.
I've occasionally plugged into a Win7 PC last year.

Not all the devices will detect they have been ejected until they are unplugged. And, with older firmware, it never happened.

And, if the device has shown the home screen, then it will be safe to disconnect the device. Waiting for the import to finish is only useful if you want to keep power to the device while it does the hard work of importing the books.

[NiLuJe]

It's also irrelevant to the underlying issue, which is how *nickel* itself is handling the unmount in order to export the partition over USBMS (which happens to be the *start* of the USBMS session, not the end).

The only difference the OS the device gets plugged into makes is in how it chooses to deal with the dirty bit being set.

----

That doesn't invalidate what @davidfor just detailed about the recent (-ish) behavioral change re: how the import screen can be triggered and when an unplug becomes safe .