Arch Linux

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#21301 - Mount Problem after Last Update

Attached to Project: Arch Linux
Opened by Paolo (berny99) - Monday, 18 October 2010, 13:43 GMT
Last edited by Tobias Powalowski (tpowa) - Tuesday, 14 February 2012, 14:33 GMT
Task Type Bug Report
Category Packages: Core
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture x86_64
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:

After 6 months I decided to upgrade my arch box, so I downloaded 1067Mb and all get updated. After the reebot this strange thing happen:

This is my fstab:

devpts /dev/pts devpts defaults 0 0
shm /dev/shm tmpfs nodev,nosuid 0 0

/dev/hda /media/cd auto ro,user,noauto,unhide 0 0
/dev/hdb /media/dvd auto ro,user,noauto,unhide 0 0
/dev/sr0 /media/dvd2 auto ro,user,noauto,unhide 0 0

UUID=92f3d6f8-c66e-4fdd-812f-398d8de04cc6 / xfs defaults,noatime,logbufs=8,logbsize=256k,nobarrier 0 0
UUID=080a4d7f-c63b-498e-b148-a4ae6266eda3 /boot xfs defaults,noatime,logbufs=8,logbsize=256k,nobarrier 0 0
UUID=06b2c055-3d03-4e7b-9d59-7fc4dfa3cf43 /home xfs defaults,noatime,logbufs=8,logbsize=256k,nobarrier 0 0
UUID=cb529a36-5be3-4cc2-8798-16d84cfa7090 /backup xfs defaults,noatime,logbufs=8,logbsize=256k,nobarrier 0 0
#External Disks
UUID=ffa8c147-0f95-49e2-a3d2-3e7ada456656 /backup2 xfs user,noauto,defaults,noatime,logbufs=8,logbsize=256k 0 0
UUID=5082322c-0cec-4985-89ba-3c1ff72ae524 /backup3 xfs user,noauto,defaults,noatime,logbufs=8,logbsize=256k 0 0
UUID=165e0dd6-87c9-4a0d-b228-df2aa65d94d9 swap swap 0 0

Please note the mount options of the root partition /
that are: defaults,noatime,logbufs=8,logbsize=256k,nobarrier
And now this is the output of the mount command:

proc on /proc type proc (rw,relatime)
sys on /sys type sysfs (rw,relatime)
udev on /dev type tmpfs (rw,nosuid,relatime,size=10240k,mode=755)
/dev/disk/by-uuid/92f3d6f8-c66e-4fdd-812f-398d8de04cc6 on / type xfs (rw,noatime,attr2,nobarrier,noquota)
fusectl on /sys/fs/fuse/connections type fusectl (rw,relatime)
devpts on /dev/pts type devpts (rw)
shm on /dev/shm type tmpfs (rw,nosuid,nodev)
/dev/sda1 on /boot type xfs (rw,noatime,logbufs=8,logbsize=256k,nobarrier)
/dev/sda4 on /home type xfs (rw,noatime,logbufs=8,logbsize=256k,nobarrier)
/dev/sdb1 on /backup type xfs (rw,noatime,logbufs=8,logbsize=256k,nobarrier)

You can see that the options are different but only for the root partition and we can find options not present at all in the fstab like attr2 and noquota.
I really don't know what is happening here, what I know is that it is not a kernel issue because I tryed with different kernel (custom compiled, arch, arch-lts).
If I boot from rescuecd I can mount the root partition and the options show up fine.
I repeated three time the test using a backup, before the update all was fine, after the system upgrade I got this strange thing on the root partition every time. All the update went fine and the system is working apparently good but the root fs is not taking the optimization options and is a little slow expecially while deleting a lot of files.
I don't know what package is causing this, may be mount or init that all got updated, there were so much packages that got updated that I can't identify but only guess.

Thank you very much for you time and help!
Regards,
Paolo Bernard

This task depends upon

Closed by  Tobias Powalowski (tpowa)
Tuesday, 14 February 2012, 14:33 GMT
Reason for closing:  Fixed
Comment by Roman Kyrylych (Romashka) - Thursday, 04 November 2010, 21:46 GMT
_Please_ never use "after last update".
It makes no sense because it's not possible to tell from which version to which every package was updated.
At least providing pacman.log is needed.

If you don't know the reason of the issue - it is better to ask on forums first instead of reporting it straight to the bugtracker
(developers' time is quite limited and there are not many developers, so there is a little chance that someone will investigate the issue,
there are much more users on forums, so your chances are much higher there)

My guess here is that it may be related to the move to busybox-based initramfs.
You may want to mount any of your non-root xfs partitions with busybox's mount command to see if it results in the same issue.
Comment by Paolo (berny99) - Friday, 05 November 2010, 09:24 GMT
Hi, Thank you for the kindly answer!
First let me apologise about the manner of the post may be not so polite for this bug tracking system.
I already put this same message on the arch forum in the workstation part but I got no answers, also asked friends and before all trying to figure out by myself what happened.
Now I attach my pacman log, what I can say is that I have a pc with windows 7 and a very minimal installation of Arch there in Virtual box, there also I have a Virtual root partiotion formatted in XFS and the same thing happened after the update in september. There I do regular update su my guess is that is one of the update of september that broke the things i think only for xfs. I will do the busybox test, so do you think that is a bug of busybox and xfs?
Any suggestion how to resolve the two month issue?
Thanks again!
Best Regards, Paolo
   pacman.txt.zip (10.9 KiB)
Comment by Paolo (berny99) - Friday, 05 November 2010, 10:56 GMT
Hi again, I tried busybox mount and I think it can be the cause. If I mount one of the non root partition with it I got
sudo busybox mount /dev/sdb1 /backup --> /dev/sdb1 on /backup type xfs (rw,relatime,attr2,noquota)

If i put some options I get:
sudo busybox mount -o defaults,noatime,logbufs=8,logbsize=256k /dev/sdb1 /backup -->
---> /dev/sdb1 on /backup type xfs (rw,noatime,attr2,logbufs=8,logbsize=256k,noquota)

So it is getting the options then adding some of it's own.
For the root I get --> /dev/disk/by-uuid/5b72d7d2-5a38-43b9-9683-2209b320368a on / type xfs (rw,noatime,attr2,noquota)

So for the root is like it is ignoring some mount options, it took noatime but it didn't took logbuf nor logbsize and added attr2 and noquota, that always.
That is my analysis hope it help.

Thanks, Paolo
Comment by Roman Kyrylych (Romashka) - Friday, 05 November 2010, 13:16 GMT
@Thomas: what do you think about this?
Comment by Thomas Bächler (brain0) - Friday, 05 November 2010, 14:50 GMT
Mounting / is a bit special.

First, / is mounted in initramfs, using busybox's mount and no options (except ro or rw), unless rootopts=xyz is on the command line. Later, / is remounted in initscripts. This remount should respect fstab and correct the mount options! This doesn't work with all file systems and all options.

This problem seems to be specific to XFS, I can try to set up a VM with XFS to investigate this. It seems that attr2 and noquota are default options for XFS, and the remount fails to change the other options appropriately.

Try working around this by adding rootopts=noatime,logbufs=8,logbsize=256k,nobarrier to your command line in the bootloader.
Comment by Paolo (berny99) - Friday, 05 November 2010, 16:21 GMT
Hi, many thanks for your help. Yes I think it is Xfs specific; anyway I tried like you said Thomas but no way, even putting that on the command line it didn't take the options. Only rw,noatime,attr2,noquota like always.

Thanks again to all of you.
Paolo
Comment by Paolo (berny99) - Monday, 08 November 2010, 10:08 GMT
Hi Thomas, I did a search and may be you got confused: the rootopts= option is for the Solaris/opensolaris OS, for linux I need to use rootflags= boot option and of course now it worked except for the added one (noquota and attr2) that of course are still there but it got all the other options.
/dev/disk/by-uuid/647384fd-65ee-4fb9-a58a-605ea95b4919 on / type xfs (rw,noatime,attr2,nobarrier,logbufs=8,logbsize=256k,noquota)
For others non root:
/dev/sda1 on /boot type xfs (rw,noatime,logbufs=8,logbsize=256k,nobarrier)

Thanks, Paolo
Comment by Greg (dolby) - Friday, 04 March 2011, 02:23 GMT
Is this still a problem?
Comment by Paolo (berny99) - Friday, 04 March 2011, 09:32 GMT
Yes, it is. The same things are happening after all this time and updates. Also I verified that if you build a custom kernel without initrd the problem is still there. I tried installing a new system in a VM and also there formatting in xfs the same happened.
Recently I also tried reiser4 and a similar thing is happening also there, where in fstab I have only noatime options inserted for reiser4 the mount command return:
(rw,noatime,atom_max_size=0x7ced3,atom_max_age=0x927c0,atom_min_size=0x100,atom_max_flushers=0x1,cbk_cache_slots=0x10)
In a fashion similar to that of XFS. I didn't try other fs so I can't say anything about that. I will try instlling gentoo also on XFS and see if there the problem also appear.
Thank you.
Regards, Paolo

Loading...