On 03/04/2016 02:54 PM, Rick Stevens wrote:
> On 03/04/2016 01:24 PM, jd1008 wrote:
>>
>>
>> On 03/04/2016 01:59 PM, Joe Zeff wrote:
>>> On 03/04/2016 12:45 PM, jd1008 wrote:
>>>> but as you know, fsck runs automagically when the system detects
>>>> insertion
>>>> into the usb port a storage device, and attempts to mount it, and find
>>>> it was
>>>> not shutdown cleanly, and runs the fsck.
>>>> So, not sure about the "grip" :)
>>>>
>>>
>>> I don't think that this is what's happening, because the system
>>> shouldn't be able to mount the drive until fsck is finished. It's
>>> always best to run fsck when the partition isn't mounted, especially
>>> if you expect it to need repair and I'd expect mount to fail if you
>>> try to run mount something that's getting fscked.
>>>
>>>> I wonder if I should open a bug at
bugzilla.redhat.com
>>>> and/or at
bugzilla.kernel.org.
>>>
>>> Next time this happens, find out if fsck is running. Run this:
>>>
>>> ps aux | grep fsck | grep -v grep
>>>
>>> in a terminal (no need for root) and if you get no response except a
>>> prompt, fsck isn't running somewhere.
>> Thanx Joe. I just replied to Rick. It was my fault, but my fault was
>> ably assisted by the
>> system not popping a banner saying the the device was not unmounted
>> cleanly. The message only
>> went into the message buffer, and never appears on the GUI file browser.
>> And unfortunately,
>> no automagic fsck was running.
>
> I think the fsck clears the "checkedDate" field in the volume header
> when it starts, indicating that an fsck is currently being done. When
> the fsck completes, it fills in the field. So, if you yank it before
> the fsck is complete, the field is clear.
>
> When you plug it in again, the system sees that it's "not clean"
> because the field is empty, but assumes that an fsck is in progress
> because the field is indeed clear (and truthfully, there was one in
> progress before you rudely interrupted it). Since it assumes there's
> an fsck in progress, it doesn't launch a new one.
>
> That's just a wild guess. Other filesystems do similar things (generally
> by setting an "in-fsck" flag in the header when starting and clearing
> it when done).
All well and good. But I wish there were a dismissable banner on the
desktop notifying
the user of this.
It's interesting. The "fsck.hfsplus" program has a "-g" flag
that's
supposed to output its messages in "GUI format", whatever the hell that
is:
-g Causes fsck.hfsplus to generate its output strings in
GUI format. This option is used when another
application with a graphical user interface (like Mac
OS X Disk Utility) is invoking the fsck.hfsplus tool.
You'd think the system would use that or something similar. Perhaps they
aren't compatible (Apple does lots of odd things and the HFS+ filesystem
is a pretty damned odd fish). Torvalds thinks it's hideous.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- "I understand Windows 2000 has a Y2K problem." -
----------------------------------------------------------------------