On 01/13/2015 07:49 PM, Jens Axboe wrote:
> On 01/12/2015 04:50 PM, Jens Axboe wrote:
>> On 01/09/2015 09:04 AM, George T Seese wrote:
>>> fio team,
>>>
>>> I've found a bug that results in fio hanging at the end of a workload
>>> involving log generation, in this case specific to latency log
>>> generation. The bug appears in version 2.1.14, 2.2.4, and possibly
>>> others. I have tested these scripts with 2.0.9 and they work. Slightly
>>> different scripts work on 2.0.14 as well.
>>>
>>> Error message and script file are copy-pasted below. The job that
>>> times out is not always the same job but, usually at least one does
>>> time out, it's worse with more targets.
>>>
>>> fio: pid=37182, got signal=11
>>> fio: job '/dev/sdc' hasn't exited in 60 seconds, it appears to be
>>> stuck. Doing forceful exit of this job.
>>>
>>> [global]
>>> bs=4k
>>> runtime=1800
>>> rw=randread
>>> ;thread=1
>>> direct=1
>>> ioengine=libaio
>>> ;offset=0
>>> ;randommap=1
>>> ;time_based=1
>>> bwavgtime=5000
>>> ;write_lat_log
>>> ;stonewall=1
>>> iodepth=1
>>>
>>> ;TargetDriveMarker
>>> [/dev/sdc]
>>> filename=/dev/sdc
>>> write_lat_log=sdc
>>>
>>> [/dev/sdd]
>>> filename=/dev/sdd
>>> write_lat_log=sdd
>>
>> Thanks, I'll take a look at this.
>
> Can you try the attached patch?

This one should be more complete (though the first one should also work 
for you, but try this one instead if you haven't used the first one).

-- 
Jens Axboe