fsck exits with error status, even though no errors are found

[JD]

On 09/26/2011 06:49 PM, Rick Stevens wrote:
> On 09/26/2011 06:29 PM, JD wrote:
>> On 09/26/2011 06:07 PM, Rick Stevens wrote:
>>> On 09/26/2011 05:55 PM, JD wrote:
>>>> On 09/26/2011 01:49 PM, Rick Stevens wrote:
>>>>> On 09/26/2011 01:25 PM, JD wrote:
>>>>>> On 09/26/2011 12:09 PM, Rick Stevens wrote:
>>>>>>> On 09/26/2011 11:59 AM, JD wrote:
>>>>>>>> kernel-2.6.35.14-96.fc14.i686
>>>>>>>>
>>>>>>>> During boot, when the time comes for fsck'ing
>>>>>>>> the file systems, whatever script is doing that,
>>>>>>>> is exiting with an error status, even though no
>>>>>>>> errors are displayed, and I am prompted to either
>>>>>>>> enter the root password, or type Contrl-D to continue.
>>>>>>>> Cntrl-D simply reboots. Entering the root password,
>>>>>>>> and running fsck manually to check all filesystems in fstab,
>>>>>>>> yields that all is well, no errors are found, and the exit
>>>>>>>> status is 0.
>>>>>>>>
>>>>>>>> Would appreciate some info on identifying the script that
>>>>>>>> does the fsck during boot.
>>>>>>> /etc/rc.d/rc.sysinit is the guy and it'll force an fsck if it sees
>>>>>>> a file called "/forcefsck" or "/.autofsck" in the root of the
>>>>>>> filesystem or if there's a "forcefsck" on the command line of the kernel
>>>>>>> (check your /etc/grub/grub.conf file).
>>>>>> Thanks Rick.
>>>>>>
>>>>>> I checked /etc/rc.d/rc.sysinit
>>>>>> and I see that it does check for the presence of files like:
>>>>>>
>>>>>> if [ -f /fsckoptions ]
>>>>>>
>>>>>>
>>>>>> if [ -f /forcefsck ]
>>>>>>
>>>>>>
>>>>>> elif [ -f /.autofsck ]
>>>>> Remember you need the "-a" option to ls to see files that begin with a
>>>>> dot, e.g. "ls -a /.autofsck".  Just making sure.
>>>>>
>>>>>> [ -f /etc/sysconfig/autofsck ]
>>>>>>
>>>>>> and I have none of these files.
>>>>>>
>>>>>> I checked /boot/grub/grub.conf and I see
>>>>>> no presence of any string like fsck or force
>>>>>> or auto in it.
>>>>>>
>>>>>> The only script I found that invokes /etc/rc.d/rc.sysinit is
>>>>>> /etc/init/rcS.conf, and it is not passing any args to it.
>>>>>>
>>>>>> I wounder if this maybe a bash problem?
>>>>> Do you have other filesystems on other partitions that might be
>>>>> triggering this?  Check your /etc/fstab file and see if any entries
>>>>> have stuff other than "0" as the last field.  Generally, "/" should
>>>>> have a "1" as the last field, "/boot" should have a "2", the rest (if
>>>>> any) should have "0".
>>>>>
>>>>> Also note that the system may force an fsck if you've exceeded the
>>>>> "mounts between fsck runs" or "interval-between-checks" set on ext2/3/4
>>>>> filesystems (and others, I think) via the "tune2fs -c" or "tune2fs -i"
>>>>> commands.  You could run "tune2fs -l" on the block device holding your
>>>>> root filesystem to see what values are set currently.
>>>>>
>>>>> Just an idea.
>>>>> ----------------------------------------------------------------------
>>>>> - Rick Stevens, Systems Engineer, C2 Hosting          ricks at nerd.com -
>>>>> - AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
>>>>> -                                                                    -
>>>>> -    First Law of Work:                                              -
>>>>> -    If you can't get it done in the first 24 hours, work nights.    -
>>>>> ----------------------------------------------------------------------
>>>> If this helps any, I instrumented /etc/rc.d/rc.sysinit
>>>> and added to it to print the full fsck command being issued
>>>> and the value of the exit status of fsck.
>>>> Well, here's what my instrumentation printed:
>>>>
>>>> fsck -T -t noopts=_netdev -A $fsckoptions<<<<<
>>>> rc = 16<<<<<
>>>>
>>>> return value of 16? And yet no fsck problems of any kind??
>>>>
>>>> So is this an fsck bug??
>>>> Has anyone else come across this?
>>> Error 16 is "EBUSY", which leads one to believe that SOMETHING that's
>>> being fsck'd isn't there or hasn't spun up or something.
>>>
>>>> ------------------ excerpt from /etc/rc.d/rc.sysinit--------------------
>>>> .
>>>> .
>>>> .
>>>>
>>>> if [ -z "$fastboot" -a "$READONLY" != "yes" ]; then
>>>>
>>>>            STRING=$"Checking filesystems"
>>>>        echo $STRING
>>>>        fsck -T -t noopts=_netdev -A $fsckoptions
>>>>        rc=$?
>>>>
>>>>        if [ "$rc" -eq "0" ]; then
>>>>            success "$STRING"
>>>>            echo
>>>>        elif [ "$rc" -eq "1" ]; then
>>>>                passed "$STRING"
>>>>            echo
>>>>        elif [ "$rc" -eq "2" -o "$rc" -eq "3" ]; then
>>>>            echo $"Unmounting file systems"
>>>>            umount -a
>>>>            mount -n -o remount,ro /
>>>>            echo $"Automatic reboot in progress."
>>>>            reboot -f
>>>>            fi
>>>>
>>>>            # A return of 4 or higher means there were serious problems.
>>>>        if [ $rc -gt 1 ]; then
>>>>            [ -n "$PLYMOUTH" ]&&   plymouth --hide-splash
>>>>
>>>>            failure "$STRING"
>>>>            echo
>>>>            echo
>>>>            echo $"*** An error occurred during the file system check."
>>>>            echo $"*** Dropping you to a shell; the system will reboot"
>>>>            echo $"*** when you leave the shell."
>>>>
>>>>                    str=$"(Repair filesystem)"
>>>>            PS1="$str \# # "; export PS1
>>>>            [ "$SELINUX_STATE" = "1" ]&&   disable_selinux
>>>>            sulogin
>>>>
>>>>            echo $"Unmounting file systems"
>>>>            umount -a
>>>>            mount -n -o remount,ro /
>>>>            echo $"Automatic reboot in progress."
>>>>            reboot -f
>>>>        elif [ "$rc" -eq "1" ]; then
>>>>            _RUN_QUOTACHECK=1
>>>>        fi
>>>> fi
>>>>
>> Well, this is a new behavior.
>> I only have one internal disk (dual boot)
>> and one external disk (5 GPT partitions),
>> and when any of those partitions is being fsck'ed
>> at boot, fsck says it is in good shape.
>> I do not see anything about a partition being
>> either busy or already opened or already mounted.
>> In fact, when I am prompted to enter control-D
>> or provide root password, I type the password
>> and run
>> mount
>> and there are NO partitions rom /dev/sdb mounted at all.
>> Only / is mounted RO.
>>
>> Furthermore, I manually run
>> /etc/rc.d/rc.sysinit
>>
>> and it passes with flying colors.
>>
>> So, something's going on at boot time that
>> is not being encountered when the script
>> is run manually.
>> Also, I do not powerdown the external drive,
>> and neither the internal drive (obviously),
>> and I simply type reboot.
>> At reboot, the same scenario of failed fsck
>> with errno 16.
> My bad.  EBUSY is errno 16, but according to fsck's info page, a return
> code of 16 is "Usage or syntax error", so it's not something not
> spinning up, but something in the command line is fubar'd.  You might
> try adding
>
> 	echo "fsck -T -t noopts=_netdev -A $fsckoptions"
>
> just before the actual call to fsck to see exactly how fsck is being
> called.  I suspect something's rotten in the $fsckoptions parameter.
Hi Rick,
I modified the fsck command (After I saved the original script),
t0

fsck -p

That's it! No other options.

fsck man page says that the -a option is mapped onto the -p option,
and is there only for legacy compatibility.

And fsck still exits with errno 16 !!