[lustre-discuss] Synchronous writes on a loaded ZFS OST

[Hans Henrik Happe]

In general Lustre is very stable. Metadata performance feels okay and we
only have one mdt on 6 SSDs (3-way mirror).

We had another issue that also are ZIL related:

http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/2016-May/013500.html

Cheers,
Hans Henrik



On 08-05-2018 21:21, Riccardo Veraldi wrote:
> I Was considering Lustre for my home dirs but I am quite frightened to
> see you have problems.
> How is the overall performance are you happy ?
> thanks
> 
> Rick
> 
> 
> On 5/8/18 5:56 AM, Hans Henrik Happe wrote:
>> Hi,
>>
>> We had some users experiencing slow vim (the editor) updates on our
>> Lustre homedirs. Turns out vim is doing some fsyncs that does not play
>> well with a loaded ZFS OST.
>>
>> We tried testing with ioping, which does synced writes (like dd with
>> conv=fdatasync). When an OST is loaded (i.e. scrubbing) the ioping time
>> is multiple seconds (5-10). Without load we get 100-300ms, which still
>> is far from what a ZFS fs can deliver.
>>
>> To test if a ZFS fs also would be affected we created a test fs on the
>> OST pool and ran ioping*. With or without a scrub running, the ping
>> times averaged at around 40ms.
>>
>> Has anyone else experienced this? Can it be helped?
>>
>> Cheers,
>> Hans Henrik
>>
>> * Used -WWW because ioping -W runs on an unlinked file and ZFS will not
>> sync those to disk.
>> _______________________________________________
>> lustre-discuss mailing list
>> lustre-discuss at lists.lustre.org
>> http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
> 
>