Solved fsck - Unexpected Soft Update Inconsistency?

Hello,
can anyone tell me what error this is:

1) when stopping dovecot the system shut down.
error-freeBSD-10-panic.jpg
error-freeBSD-10-panic.jpg



2) Try to clean the filesystem, but can not fixed it
fsck -t ufs -fy /dev/ad0s1d
error-freeBSD-10-fsck.jpg


Unexpected Soft Update Inconsistency ???
there is a directory named login in /var/run/dovecote/login it hold some sockets for dovecot
("dovecot" is an imap server I use)

What can I do to fix this?
 
A similar problem was reported and fixed in the freebsd-fs mailing list.

Thanks, I'm sure your answer is going to resolve the OP issue, but I have a question about:
isn't the message "FILESYSTEM IS CLEAN" (and more important the related filesystem flag) a bit misleading and inappropriate in this case ?

I mean, a corrupted / unclean filesystem should be prevented to be mounted automatically, and that usually is achieved by looking at "clean" flag of the filesystem.
 
Thanks, I'm sure your answer is going to resolve the OP issue, but I have a question about:
isn't the message "FILESYSTEM IS CLEAN" (and more important the related filesystem flag) a bit misleading and inappropriate in this case ?

I mean, a corrupted / unclean filesystem should be prevented to be mounted automatically, and that usually is achieved by looking at "clean" flag of the filesystem.

Sure! A dirty filesystem should not be mounted until it's been fully analysed/scanned by fsck(8). Furthermore, to avoid such scenarios the background_fsck should be disabled by default.
 
Sure! A dirty filesystem should not be mounted until it's been fully analysed/scanned by fsck(8). Furthermore, to avoid such scenarios the background_fsck should be disabled by default.
Fully agreed, but it appear it is not the case from the OP report: fsck was run but the inconsistency was not fixed, yet the filesystem is marked as clean. (most likely left as it was).

I also think that the panic occur at shutdown exactly because the filesystem was already inconsistent, HL1234 please confirm if that was the case.

EDIT: the panic occurred upon stopping dovecot.

I'm inclined to think that fsck shoud force the clean/dirty flag to "dirty" in cases where inconsistencies were not resolved.
 
I also think that the panic occur at shutdown exactly because the filesystem was already inconsistent, HL1234 please confirm if that was the case.
Yes, this was the case.
When I do stopping dovecote the panic message occurs and the system is rebooting automatically in 15 sec. First is was rebooting in an endless loop, because everytime when dovecot started it crashed. I stop running dovecot in rc.conf - then go to single user mode and make a fsck over all partition. It seems that this has fixed that. Starting was no fine, but when stop dovecot it crashed again.
It has been shown me, that there is something wrong with the directory
/var/run/dovecot/... so I decide to make better a fsck again. fsck has repair again some faults but some still not. But at all, it says FILESYSTEM CLEAN, so I try again start and stop dovecot. And dovecot also crashed again. At this point I think I has to repair the fault on disk which was not fixed and to do that I have to ask somebody because fsck doesn't work in that situation.

Thank You for helping me. I'll try the solution written in the link you give.

In between I have an another problem. I think I woud better to make an update to dovecote. pkg does no more work and the ports do not compile because the OS FreeBSD 10.0 p-18 is to old. I try to upgrade to FreeBSD version 11.0 and fetch the sources. But there are compiling has errors...
 
At this point I think I has to repair the fault on disk which was not fixed and to do that I have to ask somebody because fsck doesn't work in that situation.

Don't take this as the exact solution, just hopefully a push in the right direction:

I once made a thoughtless mistake of booting live into my FreeBSD installation while it was in 'savestate' status in VirtualBox (rawvmdk), which of course resulted in a corrupted journal, kernel panics, the 'FILESYSTEM IS CLEAN' messages after the automatic fscks, and the endless reboot cycle. My fix was to delete the mangled inode using this guide:

http://phaq.phunsites.net/2007/07/01/ufs_dirbad-panic-with-mangled-entries-in-ufs/

I ended up with the forewarned data loss, but I was at least able to boot live again and restore the deleted directory from backups.

I haven't looked into it, but that debugger tool might have alternative solutions to deleting. And since you already know which file (the dovecot binary or dir) is causing the crashes, you can breeze through most of that guide fairly quickly.

Edit: here's some more info on the 'soft updates' messages you're seeing:
11.10.2. Soft Updates
https://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/configtuning-disk.html
 
Glad to hear you got it working again. :)

I too chose to export that guide in PDF form when I first found it. I'm not sure why it's not working for you. I used Firefox on a laptop running Linux.

Edit: here you go (attached).
 

Attachments

  • ufs_dirbad-panic-with-mangled-entries-in-ufs.pdf
    9.3 KB · Views: 630
Back
Top