<div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">Am Di., 5. Nov. 2019 um 12:06 Uhr schrieb RAFI KC <<a href="mailto:rkavunga@redhat.com">rkavunga@redhat.com</a>>:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div bgcolor="#FFFFFF">
<p><br>
</p>
<div>On 11/4/19 8:46 PM, David Spisla wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div dir="ltr">
<div dir="ltr">
<div>Dear Gluster Community,</div>
<div><br>
</div>
<div>I also have a issue concerning performance. The last
days I updated our test cluster from GlusterFS v5.5 to
v7.0 . The setup in general:</div>
<div><br>
</div>
<div>2 HP DL380 Servers with 10Gbit NICs, 1
Distribute-Replica 2 Volume with 2 Replica Pairs. Client
is SMB Samba (access via vfs_glusterfs) . I did several
tests to ensure that Samba don't causes the fall.</div>
<div>The setup ist completely the same except the Gluster
Version<br>
</div>
<div>Here are my results:</div>
<div>64KiB 1MiB 10MiB
(Filesize)</div>
<div>
<div>
<div>3,49
47,41 300,50 (Values in
MiB/s with GlusterFS v5.5) <br>
</div>
<div>0,16
2,61 76,63 (Values
in MiB/s with GlusterFS v7.0) <br>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
<p><br>
</p>
<p>Can you please share the profile information [1] for both
versions? Also it would be really helpful if you can mention the
io patterns that used for this tests.<br>
</p>
<p>[1] :
<a href="https://docs.gluster.org/en/latest/Administrator%20Guide/Monitoring%20Workload/" target="_blank">https://docs.gluster.org/en/latest/Administrator%20Guide/Monitoring%20Workload/</a></p></div></blockquote><div>Hello Rafi,<br></div><div>thank you for your help.</div><div> <br></div><div>* First more information about the io patterns: As a client we use a DL360 Windws Server 2017 machine with 10Gbit NIC connected to the storage machines. The share will be mounted via SMB and the tests writes with fio. We use this job files (see attachment). Each job file will be executed separetely and there is a sleep about 60s between each test run to calm down the system before starting a new test.</div><div><br></div><div>* Attached below you find the profile output from the tests with v5.5 (ctime enabled), v7.0 (ctime enabled).</div><div><br></div><div>
<div>* Beside of the tests with Samba I did also some fio tests directly on the FUSE Mounts (locally on one of the storage nodes). The results show that there is only a small decrease of performance between v5.5 and v7.0<br></div><div>(All values in MiB/s)<br></div><div>
64KiB 1MiB 10MiB<br>
50,09 679,96 1023,02 (v5.5)<br>
47,00 656,46 977,60 (v7.0)<br></div><div><br></div><div>It seems to be that the combination of samba + gluster7.0 has a lot of problems, or not?</div>
</div><div><br>
</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div bgcolor="#FFFFFF"><p><br>
</p>
<blockquote type="cite">
<div dir="ltr">
<div dir="ltr">
<div dir="ltr">
<div>
<div>
<div><br>
</div>
<div>We
use this volume options (GlusterFS 7.0):</div>
<div><br>
</div>
<div>Volume
Name: archive1<br>
Type: Distributed-Replicate<br>
Volume ID: 44c17844-0bd4-4ca2-98d8-a1474add790c<br>
Status: Started<br>
Snapshot Count: 0<br>
Number of Bricks: 2 x 2 = 4<br>
Transport-type: tcp<br>
Bricks:<br>
Brick1: fs-dl380-c1-n1:/gluster/brick1/glusterbrick<br>
Brick2: fs-dl380-c1-n2:/gluster/brick1/glusterbrick<br>
Brick3: fs-dl380-c1-n1:/gluster/brick2/glusterbrick<br>
Brick4: fs-dl380-c1-n2:/gluster/brick2/glusterbrick<br>
Options Reconfigured:<br>
performance.client-io-threads: off<br>
nfs.disable: on<br>
storage.fips-mode-rchecksum: on<br>
transport.address-family: inet<br>
user.smb: disable<br>
features.read-only: off<br>
features.worm: off<br>
features.worm-file-level: on<br>
features.retention-mode: enterprise<br>
features.default-retention-period: 120<br>
network.ping-timeout: 10<br>
features.cache-invalidation: on<br>
features.cache-invalidation-timeout: 600<br>
performance.nl-cache: on<br>
performance.nl-cache-timeout: 600<br>
client.event-threads: 32<br>
server.event-threads: 32<br>
cluster.lookup-optimize: on<br>
performance.stat-prefetch: on<br>
performance.cache-invalidation: on<br>
performance.md-cache-timeout: 600<br>
performance.cache-samba-metadata: on<br>
performance.cache-ima-xattrs: on<br>
performance.io-thread-count: 64<br>
cluster.use-compound-fops: on<br>
performance.cache-size: 512MB<br>
performance.cache-refresh-timeout: 10<br>
performance.read-ahead: off<br>
performance.write-behind-window-size: 4MB<br>
performance.write-behind: on<br>
storage.build-pgfid: on<br>
features.ctime: on<br>
cluster.quorum-type: fixed<br>
cluster.quorum-count: 1<br>
features.bitrot: on<br>
features.scrub: Active<br>
features.scrub-freq: daily<br>
</div>
</div>
</div>
<div><br>
</div>
<div>For GlusterFS 5.5 its nearly the same except the fact
that there were 2 options to enable ctime feature. <br>
</div>
</div>
</div>
</div>
</blockquote>
<p><br>
</p>
<p><br>
Ctime stores additional metadata information as an extended
attributes which sometimes exceeds the default inode size. In such
scenarios the additional xattrs won't fit into the default size.
This will result in additional blocks to be used to store xattrs
in the inide, which will effect the latency. This is purely based
on the i/o operations and the total xattrs size stored in the
inode.<br>
<br>
Is it possible for you to repeat the test by disabling ctime or
increasing the inode size to a higher value say 1024KB?<br></p></div></blockquote><div>I will do so but for today I could not finish tests with ctime disabled (or higher inode value) because it takes a lot of time with v7.0 due to the low performance and I will perform it tomorrow. As soon as possible I give you the results.</div><div>By the way: You really mean inode size on xfs layer 1024KB? Or do you mean 1024Bytes? We use per default 512Bytes, because this is the recommended size until now . But it seems to be that there is a need for a new recommendation when using ctime feature as a default. I can not image that this is the real cause for the low performance because in v5.5 we also use ctime feature with inode size 512Bytes.<br></div><div><br></div><div>Regards</div><div>David<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div bgcolor="#FFFFFF"><p>
</p>
<p><br>
</p>
<blockquote type="cite">
<div dir="ltr">
<div dir="ltr">
<div dir="ltr">
<div>Our optimization for Samba looks like this (for every
version):</div>
<div><br>
</div>
<div>[global]<br>
workgroup = SAMBA<br>
netbios name = CLUSTER<br>
kernel share modes = no<br>
aio read size = 1<br>
aio write size = 1<br>
kernel oplocks = no<br>
max open files = 100000<br>
nt acl support = no<br>
security = user<br>
server min protocol = SMB2<br>
store dos attributes = no<br>
strict locking = no<br>
full_audit:failure = pwrite_send pwrite_recv pwrite
offload_write_send offload_write_recv create_file open
unlink connect disconnect rename chown fchown lchown chmod
fchmod mkdir rmdir ntimes ftruncate fallocate <br>
full_audit:success = pwrite_send pwrite_recv pwrite
offload_write_send offload_write_recv create_file open
unlink connect disconnect rename chown fchown lchown chmod
fchmod mkdir rmdir ntimes ftruncate fallocate <br>
full_audit:facility = local5<br>
durable handles = yes<br>
posix locking = no<br>
log level = 2<br>
max log size = 100000<br>
debug pid = yes<br>
</div>
<div><br>
</div>
<div>What can be the cause for this rapid falling of the
performance for small files? Are some of our vol options
not recommended anymore? <br>
</div>
<div>There were some patches concerning performance for
small files in v6.0 und v7.0 :<br>
</div>
<div>
<p class="MsoNormal" style="margin:0cm 0cm 0.0001pt;font-size:11pt;font-family:"Calibri",sans-serif"><span style="font-size:9pt;font-family:"Verdana",sans-serif"><a href="https://bugzilla.redhat.com/1670031" style="color:blue;text-decoration:underline" target="_blank"><span style="font-size:12pt;font-family:"\000026quot",serif;color:rgb(63,81,181);text-decoration:none">#1670031</span></a></span><span style="font-size:12pt;font-family:"Helvetica",sans-serif;background:white none repeat scroll 0% 0%"><span style="color:rgba(0,0,0,0.87);float:none;word-spacing:0px">:
performance regression seen with smallfile
workload tests</span></span><span style="font-size:9pt;font-family:"Verdana",sans-serif"><span></span></span></p>
</div>
<div>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif"><span style="font-size:12pt;font-family:"\000026quot",serif"><a href="https://bugzilla.redhat.com/1659327" style="color:blue;text-decoration:underline" target="_blank"><span style="color:rgb(63,81,181)">#1659327</span></a></span><span style="color:rgba(0,0,0,0.87);float:none;word-spacing:0px"><span style="font-size:12pt;font-family:"Helvetica",sans-serif;background:white none repeat scroll 0% 0%">: 43% regression in
small-file sequential read performance</span></span></p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif">And
one patch for the io-cache:</p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif">
</p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif"><span style="font-size:12pt;font-family:"\000026quot",serif"><a href="https://bugzilla.redhat.com/1659869" style="color:blue;text-decoration:underline" target="_blank"><span style="color:rgb(63,81,181)">#1659869</span></a></span><span style="color:rgba(0,0,0,0.87);float:none;word-spacing:0px"><span style="font-size:12pt;font-family:"Helvetica",sans-serif;background:white none repeat scroll 0% 0%">: improvements to io-cache</span></span></p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif">Regards</p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif">David
Spisla<br>
</p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif"><span style="color:rgba(0,0,0,0.87);float:none;word-spacing:0px"><span style="font-size:12pt;font-family:"Helvetica",sans-serif;background:white none repeat scroll 0% 0%"></span></span><span style="font-size:12pt;font-family:"\000026quot",serif"><span></span></span></p>
<p style="margin:12pt 0cm;font-size:11pt;font-family:"Calibri",sans-serif"><span style="color:rgba(0,0,0,0.87);float:none;word-spacing:0px"><span style="font-size:12pt;font-family:"Helvetica",sans-serif;background:white none repeat scroll 0% 0%"></span></span><span style="font-size:12pt;font-family:"\000026quot",serif"><span></span></span></p>
</div>
</div>
</div>
</div>
<br>
<fieldset></fieldset>
<pre>________
Community Meeting Calendar:
APAC Schedule -
Every 2nd and 4th Tuesday at 11:30 AM IST
Bridge: <a href="https://bluejeans.com/118564314" target="_blank">https://bluejeans.com/118564314</a>
NA/EMEA Schedule -
Every 1st and 3rd Tuesday at 01:00 PM EDT
Bridge: <a href="https://bluejeans.com/118564314" target="_blank">https://bluejeans.com/118564314</a>
Gluster-users mailing list
<a href="mailto:Gluster-users@gluster.org" target="_blank">Gluster-users@gluster.org</a>
<a href="https://lists.gluster.org/mailman/listinfo/gluster-users" target="_blank">https://lists.gluster.org/mailman/listinfo/gluster-users</a>
</pre>
</blockquote>
</div>
</blockquote></div></div>