On 10/31/18 2:05 PM, Stephen Morris wrote:
On 30/10/18 8:45 am, Ed Greshko wrote:
> On 10/30/18 4:46 AM, Stephen Morris wrote:
>> On 29/10/18 9:15 am, Ed Greshko wrote:
>>> On 10/29/18 5:23 AM, Stephen Morris wrote:
>>>> Hi,
>>>>
>>>> I've just noticed that at boot time that the FSTRIM service
>>>> is registering a
>>>> failure. The failure seems to be because it is trying to trim two
>>>> windows mount points
>>>> which are on the SSD, that are mounted as RO because I can't mount
>>>> them as RW due to
>>>> Microsoft functionality.
>>>>
>>>> Is there any way to configure the FSTRIM service to not
>>>> attempt to trim specific
>>>> partitions?
>>> Yes. But it may be more trouble than it is worth. I assume the
>>> only thing you're seeing
>>> is a message in the logs.
>> I noticed the message at boot time when I was monitoring the progress
>> of the boot. The
>> message that scrolled by said that the fstrim service had failed. It
>> was 'systemctl
>> status fstrim.service' that provided the details. I don't have the
>> exact message as I
>> didn't write it down but the same issue didn't occur this morning
>> presumably because the
>> trim service figured the was nothing to do. This morning when I
>> issued the systemctl
>> command it just said the service was 'inactive (dead)'.
> It will show "inactive" since it is a "one shot" process. Also,
it
> will only run once/week.
>
>> I have 4 partitions on the ssd, the windows system partition, the
>> windows drive c
>> partition, the Ubuntu boot partition and the Fedora boot partition.
>> The systemctl
>> command I issued yesterday to get the failure details indicated that
>> the Fstrim Service
>> was trying to do its work on those 4 partitions via the mount points
>> specified in fstab,
>> and, that the process on the two linux partitions was successful, but
>> the process on the
>> two windows partitions both failed.
>>
>> In fstab the two windows partitions are specified as 'Read Only'
>> because of Microsoft
>> functionality, so in my view, it is a defect in the Fstrim.service
>> processing to even
>> attempt any write processes on a mount point that is 'Read Only'. I
>> can understand the
>> functionality of "--all", but in my view a bit of common sense logic
>> needs to be
>> included with that to bypass any partitions that are 'Read Only' via
>> the methodology
>> that is being indicated it is using.
> Well, the process probably doesn't check the fstab. According to the
> man page ...
>
> Errors from filesystems that do not support the discard operation are
> silently ignored.
>
> Is all that is checked/ignored.
>
> If the messages bother you, you can bugzilla it. But it seems quite
> minor.
You are correct, it is minor, but I just don't see why, when the fstrim
process tries to trim the 4 mount points specified in fstab for the 4
partitions on the SSD, and 2 are successful and two fail, potentially
because those mount points are RO and Microsoft antics won't allow them
to be RW, that should mean that the fstrim service has failed. I can
understand the service being flagged a failing for all processes failed,
but if some work then I would have expected a warning that a subset of
what it was doing didn't work. That's the way I write my programs when I
consider them to be written properly.
Generally speaking, programs exit with a "return code" or "exit
status".
By convention, if the return code is zero, then the program has run
properly. A non-zero return code indicates a failure of some kind.
Now, there is no rule as to what _kind_ of failure a non-zero return
code means.
The man page for fstrim(8) specifically says:
0 success
1 failure
32 all failed
64 some filesystem discards have succeeded, some failed
So in your case, it's probably returning 64. AFAIK, systemd considers
ANY non-zero return code as a failure.
If it truly bothers you, ask the systemd folk to provide a means to
exclude certain return codes from "fail" status.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- Never eat anything larger than your head -
----------------------------------------------------------------------