From: "John Stoffel" <john@stoffel.org>
To: Adam Goryachev <mailinglists@websitemanagers.com.au>
Cc: linux-raid@vger.kernel.org
Subject: Re: RAID performance - new kernel results
Date: Mon, 15 Apr 2013 11:31:54 -0400 [thread overview]
Message-ID: <20844.7530.817412.979677@quad.stoffel.home> (raw)
In-Reply-To: <516BF13B.4010704@websitemanagers.com.au>
Hi Adam,
Thanks for all your posting and continued updates on your travails
about getting the performance on your Xen systems and storage. A
complete eye-opener in alot of ways.
Now, from looking over your report, it strongly smells of a problem in
the network switch. I think you have just one in the core of your
network, correct? I'd probably try to bring up a test network (if you
have the spare systems in a lab) and try to replicate the packet
drops.
But in general, I'd probably:
- remove the iSCSI bonding, goto a single 1Gb link.
- get rid of jumbo frames, if you're using it.
- can you reduce the size of your bond0 on the storage box?
I wonder if the switch is having some sort of table over-flow, or is
just having some sort of brain fart and droppping a packet and then
needs time to rebuild it's tables internally to get things going
again?
I'd try to borrow a similar sized switch from another vendor and try
using that instead if you can. Another thing is to try and use SNMP
to grab stats from the switch and look for patterns. When you see
connectivity problems, do you see a corresponding drop on one of the
links on the bond0 connection? Or on another bond?
But, thinking about it more, you don't mention if you're dropping
packets on the iSCSI side of things, or just on the regular network.
That's a key observation, since it will either suggest, or refute my
idea of the problem being in the bond(s).
Do you see any errors in the dmesg logs on the Xen/Linux/Windows
boxes? And when you have an outage between two hosts, do pings to
*other* hosts still work just fine, or does all network traffic on
that host come to a stop?
It really smells of a switch problem. Have you checked that the
switch firmware is upto date? It might just be that Netgear makes a
crappy switch (cue people to chime on on this! :-) which can't handle
the load you're tossing at it. Which is why I suggest you try another
vendor's switch.
Cisco is probably reliable but expensive. Dell has some ok switches
in my experience, but nothing recent. I've heard good things about
other brands such as Juniper, Force10 (now Dell) and others.
Please keep posting, it's great information for the rest of us to keep
in the back of our heads.
John
next prev parent reply other threads:[~2013-04-15 15:31 UTC|newest]
Thread overview: 131+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-02-07 6:48 RAID performance Adam Goryachev
2013-02-07 6:51 ` Adam Goryachev
2013-02-07 8:24 ` Stan Hoeppner
2013-02-07 7:02 ` Carsten Aulbert
2013-02-07 10:12 ` Adam Goryachev
2013-02-07 10:29 ` Carsten Aulbert
2013-02-07 10:41 ` Adam Goryachev
2013-02-07 8:11 ` Stan Hoeppner
2013-02-07 10:05 ` Adam Goryachev
2013-02-16 4:33 ` RAID performance - *Slow SSDs likely solved* Stan Hoeppner
[not found] ` <cfefe7a6-a13f-413c-9e3d-e061c68dc01b@email.android.com>
2013-02-17 5:01 ` Stan Hoeppner
2013-02-08 7:21 ` RAID performance Adam Goryachev
2013-02-08 7:37 ` Chris Murphy
2013-02-08 13:04 ` Stan Hoeppner
2013-02-07 9:07 ` Dave Cundiff
2013-02-07 10:19 ` Adam Goryachev
2013-02-07 11:07 ` Dave Cundiff
2013-02-07 12:49 ` Adam Goryachev
2013-02-07 12:53 ` Phil Turmel
2013-02-07 12:58 ` Adam Goryachev
2013-02-07 13:03 ` Phil Turmel
2013-02-07 13:08 ` Adam Goryachev
2013-02-07 13:20 ` Mikael Abrahamsson
2013-02-07 22:03 ` Chris Murphy
2013-02-07 23:48 ` Chris Murphy
2013-02-08 0:02 ` Chris Murphy
2013-02-08 6:25 ` Adam Goryachev
2013-02-08 7:35 ` Chris Murphy
2013-02-08 8:34 ` Chris Murphy
2013-02-08 14:31 ` Adam Goryachev
2013-02-08 14:19 ` Adam Goryachev
2013-02-08 6:15 ` Adam Goryachev
2013-02-07 15:32 ` Dave Cundiff
2013-02-08 13:58 ` Adam Goryachev
2013-02-08 21:42 ` Stan Hoeppner
2013-02-14 22:42 ` Chris Murphy
2013-02-15 1:10 ` Adam Goryachev
2013-02-15 1:40 ` Chris Murphy
2013-02-15 4:01 ` Adam Goryachev
2013-02-15 5:14 ` Chris Murphy
2013-02-15 11:10 ` Adam Goryachev
2013-02-15 23:01 ` Chris Murphy
2013-02-17 9:52 ` RAID performance - new kernel results Adam Goryachev
2013-02-18 13:20 ` RAID performance - new kernel results - 5x SSD RAID5 Stan Hoeppner
2013-02-20 17:10 ` Adam Goryachev
2013-02-21 6:04 ` Stan Hoeppner
2013-02-21 6:40 ` Adam Goryachev
2013-02-21 8:47 ` Joseph Glanville
2013-02-22 8:10 ` Stan Hoeppner
2013-02-24 20:36 ` Stan Hoeppner
2013-03-01 16:06 ` Adam Goryachev
2013-03-02 9:15 ` Stan Hoeppner
2013-03-02 17:07 ` Phil Turmel
2013-03-02 23:48 ` Stan Hoeppner
2013-03-03 2:35 ` Phil Turmel
2013-03-03 15:19 ` Adam Goryachev
2013-03-04 1:31 ` Phil Turmel
2013-03-04 9:39 ` Adam Goryachev
2013-03-04 12:41 ` Phil Turmel
2013-03-04 12:42 ` Stan Hoeppner
2013-03-04 5:25 ` Stan Hoeppner
2013-03-03 17:32 ` Adam Goryachev
2013-03-04 12:20 ` Stan Hoeppner
2013-03-04 16:26 ` Adam Goryachev
2013-03-05 9:30 ` RAID performance - 5x SSD RAID5 - effects of stripe cache sizing Stan Hoeppner
2013-03-05 15:53 ` Adam Goryachev
2013-03-07 7:36 ` Stan Hoeppner
2013-03-08 0:17 ` Adam Goryachev
2013-03-08 4:02 ` Stan Hoeppner
2013-03-08 5:57 ` Mikael Abrahamsson
2013-03-08 10:09 ` Stan Hoeppner
2013-03-08 14:11 ` Mikael Abrahamsson
2013-02-21 17:41 ` RAID performance - new kernel results - 5x SSD RAID5 David Brown
2013-02-23 6:41 ` Stan Hoeppner
2013-02-23 15:57 ` RAID performance - new kernel results John Stoffel
2013-03-01 16:10 ` Adam Goryachev
2013-03-10 15:35 ` Charles Polisher
2013-04-15 12:23 ` Adam Goryachev
2013-04-15 15:31 ` John Stoffel [this message]
2013-04-17 10:15 ` Adam Goryachev
2013-04-15 16:49 ` Roy Sigurd Karlsbakk
2013-04-15 20:16 ` Phil Turmel
2013-04-16 19:28 ` Roy Sigurd Karlsbakk
2013-04-16 21:03 ` Phil Turmel
2013-04-16 21:43 ` Stan Hoeppner
2013-04-15 20:42 ` Stan Hoeppner
2013-02-08 3:32 ` RAID performance Stan Hoeppner
2013-02-08 7:11 ` Adam Goryachev
2013-02-08 17:10 ` Stan Hoeppner
2013-02-08 18:44 ` Adam Goryachev
2013-02-09 4:09 ` Stan Hoeppner
2013-02-10 4:40 ` Adam Goryachev
2013-02-10 13:22 ` Stan Hoeppner
2013-02-10 16:16 ` Adam Goryachev
2013-02-10 17:19 ` Mikael Abrahamsson
2013-02-10 21:57 ` Adam Goryachev
2013-02-11 3:41 ` Adam Goryachev
2013-02-11 4:33 ` Mikael Abrahamsson
2013-02-12 2:46 ` Stan Hoeppner
2013-02-12 5:33 ` Adam Goryachev
2013-02-13 7:56 ` Stan Hoeppner
2013-02-13 13:48 ` Phil Turmel
2013-02-13 16:17 ` Adam Goryachev
2013-02-13 20:20 ` Adam Goryachev
2013-02-14 12:22 ` Stan Hoeppner
2013-02-15 13:31 ` Stan Hoeppner
2013-02-15 14:32 ` Adam Goryachev
2013-02-16 1:07 ` Stan Hoeppner
2013-02-16 17:19 ` Adam Goryachev
2013-02-17 1:42 ` Stan Hoeppner
2013-02-17 5:02 ` Adam Goryachev
2013-02-17 6:28 ` Stan Hoeppner
2013-02-17 8:41 ` Adam Goryachev
2013-02-17 13:58 ` Stan Hoeppner
2013-02-17 14:46 ` Adam Goryachev
2013-02-19 8:17 ` Stan Hoeppner
2013-02-20 16:45 ` Adam Goryachev
2013-02-21 0:45 ` Stan Hoeppner
2013-02-21 3:10 ` Adam Goryachev
2013-02-22 11:19 ` Stan Hoeppner
2013-02-22 15:25 ` Charles Polisher
2013-02-23 4:14 ` Stan Hoeppner
2013-02-12 7:34 ` Mikael Abrahamsson
2013-02-08 7:17 ` Adam Goryachev
2013-02-07 12:01 ` Brad Campbell
2013-02-07 12:37 ` Adam Goryachev
2013-02-07 17:12 ` Fredrik Lindgren
2013-02-08 0:00 ` Adam Goryachev
2013-02-11 19:49 ` Roy Sigurd Karlsbakk
2013-02-11 20:30 ` Dave Cundiff
2013-02-07 11:32 ` Mikael Abrahamsson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20844.7530.817412.979677@quad.stoffel.home \
--to=john@stoffel.org \
--cc=linux-raid@vger.kernel.org \
--cc=mailinglists@websitemanagers.com.au \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).