[OpenIndiana-discuss] Sudden ZFS performance issue

Saso Kiselkov skiselkov.ml at gmail.com
Sat Jul 6 08:36:01 UTC 2013


Run the commands with a time interval, not just once - that will show
you the aggregate values since boot, which if boot was a long time ago
can hide any sudden abnormalities. Also, do some dd'ing while you're
running iostat to stress the system out. If it's idle then the tests are
meaningless.

--
Saso

On 06/07/2013 04:52, wim at vandenberge.us wrote:
> Here is the output for "iostat -xn" for the smaller of the two servers/pools
> (80TB). The three c8 drives with the 0 across the board are the hot spares.
> Nothing jumps out at me. The c5 drives are boot (mirror), ZIL (mirror) and L2ARC
> (stripe)
> 
> 
> 
>     r/s    w/s   kr/s   kw/s wait actv wsvc_t asvc_t  %w  %b device
> 
>     0.1    0.8    0.1    1.6  0.0  0.0    0.0    0.3   0   0 c5t2d0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c5t0d0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c5t1d0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c5t3d0
> 
>    36.6  162.6  420.0 18840.0  0.0  0.6    0.0    3.2   0   8 c5t4d0
> 
>    51.3  160.5  719.9 18236.1  0.0  0.6    0.0    2.9   0   8 c5t5d0
> 
>   867.6    0.6 2558.1    0.4  0.0  0.9    0.0    1.0   0  29
> c8t5000C50056592673d0
> 
>   778.3    6.7 2065.3    8.1  0.0  0.7    0.0    0.9   0  28
> c8t5000C50056571453d0
> 
>   938.3    0.6 2585.0    0.4  0.0  0.7    0.0    0.7   0  24
> c8t5000C5005658F473d0
> 
>   982.4    0.6 2602.4    0.4  0.0  0.6    0.0    0.6   0  24
> c8t5000C5005658F8C3d0
> 
>   866.8    6.9 2197.2   10.0  0.0  0.7    0.0    0.8   0  27
> c8t5000C5005652C613d0
> 
>   827.9    6.5 2300.4    9.8  0.0  0.9    0.0    1.0   0  29
> c8t5000C5005655A633d0
> 
>   862.3    6.6 2290.2    9.8  0.0  0.7    0.0    0.9   0  28
> c8t5000C5005655B553d0
> 
>   868.2    6.3 2304.4    9.5  0.0  0.7    0.0    0.9   0  26
> c8t5000C500565659D3d0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0
> c8t5000C50056593873d0
> 
>   985.4    5.2 2545.8    6.5  0.0  0.7    0.0    0.7   0  25
> c8t5000C5005657D007d0
> 
>   990.3    5.8 2442.8    6.4  0.0  0.7    0.0    0.7   0  26
> c8t5000C5005657AD37d0
> 
>   963.7    5.6 2406.0    6.5  0.0  0.7    0.0    0.8   0  25
> c8t5000C5005658E2B7d0
> 
>   976.3    0.6 2627.6    0.4  0.0  0.6    0.0    0.6   0  23
> c8t5000C50056590B3Bd0
> 
>   917.4    0.6 2523.0    0.4  0.0  0.7    0.0    0.7   0  24
> c8t5000C5005658E3ABd0
> 
>   929.3    0.6 2496.6    0.4  0.0  0.8    0.0    0.9   0  30
> c8t5000C5005659283Bd0
> 
>   862.4    6.5 2189.0    9.7  0.0  0.8    0.0    0.9   0  28
> c8t5000C500565630FBd0
> 
>   938.0    0.6 2604.8    0.4  0.0  0.7    0.0    0.7   0  24
> c8t5000C5005659135Bd0
> 
>   886.1    0.6 2492.7    0.4  0.0  0.9    0.0    1.1   0  31
> c8t5000C5005659248Fd0
> 
>   955.6    5.7 2466.4    7.1  0.0  0.8    0.0    0.8   0  26
> c8t5000C500565880EFd0
> 
>   857.4    0.6 2538.8    0.4  0.0  0.9    0.0    1.1   0  31
> c8t5000C50056592ACFd0
> 
>   831.5    0.6 2461.6    0.4  0.0  1.1    0.0    1.4   0  33
> c8t5000C50056591F5Fd0
> 
>   856.8    0.6 2422.2    0.4  0.0  0.9    0.0    1.0   0  29
> c8t5000C5005659255Fd0
> 
>  1011.3    5.4 2471.2    6.5  0.0  0.7    0.0    0.7   0  25
> c8t5000C5005658E17Fd0
> 
>   975.7    5.2 2481.4    6.8  0.0  0.7    0.0    0.8   0  26
> c8t5000C5005658DB3Fd0
> 
>   859.2    6.3 2289.7    9.5  0.0  0.8    0.0    0.9   0  27
> c8t5000C50056561053d0
> 
>   779.8    6.6 1961.4    8.1  0.0  0.7    0.0    0.9   0  26
> c8t5000C50056577043d0
> 
>   917.4    0.6 2623.7    0.4  0.0  0.7    0.0    0.8   0  25
> c8t5000C5005658E6A3d0
> 
>   848.6    7.0 2163.1    9.7  0.0  0.8    0.0    0.9   0  28
> c8t5000C5005655A603d0
> 
>   741.9    7.1 1960.3    8.0  0.0  0.8    0.0    1.0   0  27
> c8t5000C50056576573d0
> 
>   786.5    5.6 1889.8    8.1  0.0  0.7    0.0    0.9   0  26
> c8t5000C5005657A2E3d0
> 
>   952.2    0.6 2629.5    0.4  0.0  0.7    0.0    0.8   0  25
> c8t5000C500565909E3d0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0
> c8t5000C50056594023d0
> 
>   988.6    5.9 2526.3    6.9  0.0  0.8    0.0    0.8   0  26
> c8t5000C5005657F3E3d0
> 
>   775.0    6.9 2067.2    8.1  0.0  0.7    0.0    1.0   0  26
> c8t5000C50056579ED7d0
> 
>   862.6    6.8 2307.5   10.0  0.0  0.9    0.0    1.0   0  29
> c8t5000C5005652AE27d0
> 
>   876.8    0.6 2497.3    0.4  0.0  1.0    0.0    1.1   0  31
> c8t5000C500565929B7d0
> 
>   976.0    5.1 2517.6    6.3  0.0  0.7    0.0    0.7   0  26
> c8t5000C5005658E25Bd0
> 
>   714.6    6.8 2000.5    8.0  0.0  0.9    0.0    1.2   0  29
> c8t5000C50056569ADBd0
> 
>   928.2    0.6 2660.8    0.4  0.0  0.7    0.0    0.8   0  24
> c8t5000C5005658EF5Bd0
> 
>   858.8    0.6 2421.0    0.4  0.0  0.9    0.0    1.1   0  30
> c8t5000C500565924CBd0
> 
>   818.0    6.3 2188.1    9.5  0.0  0.9    0.0    1.1   0  30
> c8t5000C5005656443Bd0
> 
>   904.5    0.6 2603.6    0.4  0.0  0.9    0.0    1.0   0  31
> c8t5000C5005659272Bd0
> 
>   735.2    6.4 1993.1    7.8  0.0  0.8    0.0    1.1   0  27
> c8t5000C500565776AFd0
> 
>   794.2    7.3 1995.3    8.0  0.0  0.7    0.0    0.8   0  25
> c8t5000C5005657640Fd0
> 
>   961.9    5.5 2509.9    6.6  0.0  0.7    0.0    0.7   0  26
> c8t5000C5005658E0BFd0
> 
>     0.0    0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0
> c8t5000C50056593E3Fd0
> 
>   791.0    6.9 2105.8    7.8  0.0  0.8    0.0    1.0   0  30
> c8t5000C500565696FFd0
> 
>   916.8    0.6 2637.3    0.4  0.0  0.7    0.0    0.8   0  25
> c8t5000C50056590E1Fd0
> 
> 
> 
> 
>> On July 5, 2013 at 4:27 PM Saso Kiselkov <skiselkov.ml at gmail.com> wrote:
>>
>>
>> On 05/07/2013 21:00, wim at vandenberge.us wrote:
>>> Latencytop reports the following continuously for the pool process and it
>>> doesn’t change significantly under load, which looks ok to me:
>>>
>>> genunix`cv_wait genunix`taskq_threa 14491 9.6 msec 6.2 sec 99.3 %
>>>
>>> Wait for available CPU 14618 44.9 usec 9.8 msec 0.5 %
>>>
>>> Adapt. lock spin 14501 24.2 usec 1.7 msec 0.3 %
>>>
>>> genunix`turnstile_block unix`mutex_ 48 174.5 usec 453.9 usec 0.0 %
>>>
>>> Spinlock spin 194 35.4 usec 309.0 usec 0.0 %
>>>
>>> genunix`turnstile_block unix`mutex_ 20 315.5 usec 3.1 msec 0.0 %
>>>
>>> genunix`turnstile_block unix`mutex_ 62 56.1 usec 124.5 usec 0.0 %
>>>
>>> I realize 64GB is low for this size of storage. would it be conceivable that
>>> I
>>> reached some threshold there?
>>
>> What's 'iostat -xn' showing in the asvc_t and %b columns? Isn't some
>> disk taking an excessive amount of time to fulfill I/O requests? For
>> example one of the disks in the pool showing 100% busy and very large
>> service times can slow down the whole raidz and by extension the pool
>> (all blocks read from the raidz will incur a significant hit).
>>
>> Cheers,
>> --
>> Saso
>>
>> _______________________________________________
>> OpenIndiana-discuss mailing list
>> OpenIndiana-discuss at openindiana.org
>> http://openindiana.org/mailman/listinfo/openindiana-discuss
> _______________________________________________
> OpenIndiana-discuss mailing list
> OpenIndiana-discuss at openindiana.org
> http://openindiana.org/mailman/listinfo/openindiana-discuss
> 




More information about the OpenIndiana-discuss mailing list