zfs, cam sticking on failed disk

[Steven Hartland]

On 07/05/2015 10:50, Slawa Olhovchenkov wrote:
> On Thu, May 07, 2015 at 09:41:43AM +0100, Steven Hartland wrote:
>
>> On 07/05/2015 09:07, Slawa Olhovchenkov wrote:
>>> I have zpool of 12 vdev (zmirrors).
>>> One disk in one vdev out of service and stop serving reuquest:
>>>
>>> dT: 1.036s  w: 1.000s
>>>    L(q)  ops/s    r/s   kBps   ms/r    w/s   kBps   ms/w   %busy Name
>>>       0      0      0      0    0.0      0      0    0.0    0.0| ada0
>>>       0      0      0      0    0.0      0      0    0.0    0.0| ada1
>>>       1      0      0      0    0.0      0      0    0.0    0.0| ada2
>>>       0      0      0      0    0.0      0      0    0.0    0.0| ada3
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da0
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da1
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da2
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da3
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da4
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da5
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da6
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da7
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da8
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da9
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da10
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da11
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da12
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da13
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da14
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da15
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da16
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da17
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da18
>>>      24      0      0      0    0.0      0      0    0.0    0.0| da19
>>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da20
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da21
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da22
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da23
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da24
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da25
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da26
>>>       0      0      0      0    0.0      0      0    0.0    0.0| da27
>>>
>>> As result zfs operation on this pool stoped too.
>>> `zpool list -v` don't worked.
>>> `zpool detach tank da19` don't worked.
>>> Application worked with this pool sticking in `zfs` wchan and don't killed.
>>>
>>> # camcontrol tags da19 -v
>>> (pass19:isci0:0:3:0): dev_openings  7
>>> (pass19:isci0:0:3:0): dev_active    25
>>> (pass19:isci0:0:3:0): allocated     25
>>> (pass19:isci0:0:3:0): queued        0
>>> (pass19:isci0:0:3:0): held          0
>>> (pass19:isci0:0:3:0): mintags       2
>>> (pass19:isci0:0:3:0): maxtags       255
>>>
>>> How I can cancel this 24 requst?
>>> Why this requests don't timeout (3 hours already)?
>>> How I can forced detach this disk? (I am lready try `camcontrol reset`, `camconrol rescan`).
>>> Why ZFS (or geom) don't timeout on request and don't rerouted to da18?
>>>
>> If they are in mirrors, in theory you can just pull the disk, isci will
>> report to cam and cam will report to ZFS which should all recover.
> Yes, zmirror with da18.
> I am surprise that ZFS don't use da18. All zpool fully stuck.
A single low level request can only be handled by one device, if that 
device returns an error then ZFS will use the other device, but not until.
>
>> With regards to not timing out this could be a default issue, but having
> I am understand, no universal acceptable timeout for all cases: good
> disk, good saturated disk, tape, tape library, failed disk, etc.
> In my case -- failed disk. This model already failed (other specimen)
> with same symptoms).
>
> May be exist some tricks for cancel/aborting all request in queue and
> removing disk from system?
Unlikely tbh, pulling the disk however should.
>
>> a very quick look that's not obvious in the code as
>> isci_io_request_construct etc do indeed set a timeout when
>> CAM_TIME_INFINITY hasn't been requested.
>>
>> The sysctl hw.isci.debug_level may be able to provide more information,
>> but be aware this can be spammy.
> I am already have this situation, what command interesting after
> setting hw.isci.debug_level?
I'm afraid I'm not familiar isci I'm afraid possibly someone else who is 
can chime in.

     Regards
     Steve