mmmmn

版权声明：转载时请以超链接形式标明文章原始出处和作者信息及本声明
http://mmmmn.blogbus.com/logs/2329304.html

1、Re: i/o is high

------------------------------------------------------------------------------

--

From: "Dexthor" <gmolakal@xxxxxxxxx>
Date: 29 Mar 2006 06:19:00 -0800

------------------------------------------------------------------------------

--
r/s w/s Mr/s Mw/s wait actv wsvc_t asvc_t %w %b
device
325.6 13 2.9 0.1 0 3 0 8.9 0 98
A00000039d0
151.6 12.2 1.2 0.1 0 2.7 0 16.7 0 98
A00000038d0
888.8 1.2 17.2 0 0 2.3 0 2.5 1 98
A0000038Ed0

So, 17.2 MB is being read in 888.8 requests, so your avg: I/O request
size = 19.8 KB.
Your %busy is so high because, you are maxing out the Disk Queue with
large I/O request broken down into multiple small requests.

Here's what I would do:
1. Review sd_max_throttle setting that is applicable to your Disk
Subsystem and set it. EMC recommends 20.
2. vol_maxio - set this to something reasonable and relevant to your
disk. Suppose you have a SAN and using 2xGbit HBAs, then set this to
16000 (~16MB).
3. If you have striped your volumes today, then your problem is
probably linked to your Stripe Unit size. A large sequential I/O
request is broken down into multiple small requests(of size stripe
unit) by your Volume Manager Disk driver. If that is the case, you have
to restripe your stripe set with decent stripe unit size. In a Gbit/SAN
environment it is better to have StripeUnit == vol_maxio.

And talk to your Disk Management group to see if there are any
hotspots.

-Dexthor.

.

------------------------------------------------------------------------------

--

References:
i/o is high
From: atif76
Prev by Date: Re: Looking For 10/100 Ethernet Card With Solaris 8 Drivers
Next by Date: Re: New ssh/sshd patches for Solaris 9
Previous by thread: i/o is high
Next by thread: state in single user mode
Index(es):
Date
Thread
We are proud to have Web Hosting and Rack Housing from 9 Net Avenue

Deutschland.
(03)
Security UNIX Linux Coding Shop Directory Usenet

Mailing-ListsNewsgroupsAboutPrivacySearch  unix.derkeiler.com  > Newsgroups  >

comp.unix.solaris  > 2006-03      

2、Re: solaris big crash

------------------------------------------------------------------------------

--

From: Casper H.S. Dik <Casper.Dik@xxxxxxx>
Date: 28 Mar 2006 18:08:23 GMT

------------------------------------------------------------------------------

--
globulon@xxxxxxxxx writes:

we are running under Solaris 10 x86. Our system suddenly crashed a few
days ago. Symptoms are quite bizarre.
We found first that no ssh remote connection was possible. After a
reboot it was possible again for small amount of time to connect.

You are running an unpatched system and you've been hit but the
DST bug from hell.

You will need to set the date/time correctly from Solaris and then
reboot.

If the system only runs Solaris, I suggest you also run the following
command before adjusting the date:

rtc -c -z GMT

Then apply the appropriate patches.

Casper
.

------------------------------------------------------------------------------

--

Follow-Ups:
Re: solaris big crash
From: Wes W
References:
solaris big crash
From: globulon
Prev by Date: Re: Solaris Filesystem and Power Failures
Next by Date: Re: Solaris Root Mirroring
Previous by thread: Re: solaris big crash
Next by thread: Re: solaris big crash
Index(es):
Date
Thread
We are proud to have Web Hosting and Rack Housing from 9 Net Avenue

Deutschland.
(08)
Security UNIX Linux Coding Shop Directory Usenet

Mailing-ListsNewsgroupsAboutPrivacySearch  unix.derkeiler.com  > Newsgroups  >

comp.unix.solaris  > 2006-03      
-------
Re: solaris big crash

------------------------------------------------------------------------------

--

From: Casper H.S. Dik <Casper.Dik@xxxxxxx>
Date: 29 Mar 2006 08:53:23 GMT

------------------------------------------------------------------------------

--
Wes W <wes@xxxxxxxxxxxxxxx> writes:

Casper H.S. Dik wrote:

globulon@xxxxxxxxx writes:

we are running under Solaris 10 x86. Our system suddenly crashed a few
days ago. Symptoms are quite bizarre.
We found first that no ssh remote connection was possible. After a
reboot it was possible again for small amount of time to connect.

You are running an unpatched system and you've been hit but the
DST bug from hell.

You will need to set the date/time correctly from Solaris and then
reboot.

If the system only runs Solaris, I suggest you also run the following
command before adjusting the date:

rtc -c -z GMT

Then apply the appropriate patches.

Casper

In addition to Casper's excellent insight, you may wish to join some of
Sun's free email newsletters regarding Solaris. The bug Casper
identified has been warned about or otherwise acknowledged for several
weeks at least.

Years, really; I think we warned both previous DST changes.

(the bug is limited to x64 running 64 bit kernels changing over to
timezones with a negative offset from GMT; a patch for this issue was
released in June 05)

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.
.

------------------------------------------------------------------------------

--

Follow-Ups:
Re: solaris big crash
From: globulon
References:
solaris big crash
From: globulon
Re: solaris big crash
From: Casper H . S . Dik
Re: solaris big crash
From: Wes W
Prev by Date: Re: observe packets on Unix domain socket
Next by Date: Re: Looking for an USB PCI Card for my Ultra 10
Previous by thread: Re: solaris big crash
Next by thread: Re: solaris big crash
Index(es):
Date
Thread
We are proud to have Web Hosting and Rack Housing from 9 Net Avenue

Deutschland.
(10)
Security UNIX Linux Coding Shop Directory Usenet

Mailing-ListsNewsgroupsAboutPrivacySearch  unix.derkeiler.com  > Newsgroups  >

comp.unix.solaris  > 2006-03