[lustre-discuss] Tuning for metadata performance
Michael Di Domenico
mdidomenico4 at gmail.com
Fri Jan 15 06:24:01 PST 2021
On Thu, Jan 14, 2021 at 3:34 PM Vicker, Darby J. (JSC-EG111)[Jacobs
Technology, Inc.] <darby.vicker-1 at nasa.gov> wrote:
> This took a very long time for me to clone from the web. But just cloning from disk to disk on a local SSD (git clone linux-stable linux-stable2) takes about 2 minutes - about the same as the repo I've been using. I just cloned it from the local SSD to lustre and it took me about 11.5 mintues for the clone. That timing is in line to what I reported earlier if you scale by the number of files.
>
a first blush pass a cloning the kernel produced, 4mins on nfs and
8mins on lustre. i've done nothing to check/test/etc, literally just
the git clones. the source repo is sitting on a raid10 of 4 ssd's,
the box is a server machine plenty of cpu's/memory and 10G ethernet
so at least our numbers match. my lustre system might be a little
faster or less utilized then yours. which might account for the
shorter lustre time
> I would also love to know what your "out of the box" io500 MD test numbers look like (./io500.sh config-minimal.ini) as those should be a good data to compare too.
i've had no end of trouble trying to get this test to run and gave up.
i'll give it another go and see what i can pull out. the framework
isn't the greatest in my opinion. since i can't quesce my system, i'm
not sure how relevant the results would be anyhow.
More information about the lustre-discuss
mailing list