[THIN] Re: XenServer Slowdown
- From: "Michael Monasterio" <michaelm@xxxxxxxxxxxxxx>
- To: <thin@xxxxxxxxxxxxx>
- Date: Wed, 17 Nov 2010 20:21:34 -0500
James, you might also check the process handles for all processes and
make sure no set of processes is consuming a large amount of handles.
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of Joe Shonk
Sent: Wednesday, November 17, 2010 7:35 PM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
Yeah, I thought the same thing when I typed it up. My best guess is a
memory leak or a bad driver.
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of Dan Dill
Sent: Wednesday, November 17, 2010 3:45 PM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
While it's good to nail down the memory config it seems strange that the
memory would affect the server coming to a crawl after a period of
time.... That symptom more points to software I would think...
Dan Dill
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of Berny Stapleton
Sent: Wednesday, November 17, 2010 2:21 PM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
Don't forget the toss up between UDIMM vs RDIMM.
UDIMM is faster than RDIMM but not registered, there is a memory limit
on the amount of UDIMM you can run as well. Depending on the situation
your in, this might be a trade off your willing to make.
Berny
On Wed, Nov 17, 2010 at 9:05 PM, Joe Shonk <joe.shonk@xxxxxxxxx> wrote:
32 gigs is the limit for Windows Server Standard 2008 x64. Enterprise
will do 1 terabyte.
The one thing that jumps out at me is the number 36. Generally with a 2
proc Nehalem based system you would expect to see memory in the
following multiples: 1, 12, 24, 48 or 96. Yes there are other memory
configurations that work but they generally indicate that the memory is
not optimized.
The number 36 means to me that you're running either (6 x 4gb dimms + 6
x 2gb dimms) or (18 x 2gb dimms). The later configuration would choke
your memory bus speed from 1333MHz to 800 MHz.
I would also check to see what rank the dimms you have are.
Joe
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of James Scanlon
Sent: Wednesday, November 17, 2010 9:46 AM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
XenApp - Correct
Windows 2008 SP2 - not R2,
Windows system reports 36GB in system properties, HP SIM reports 32GB of
ram addressable / available - WTF? - chasing this up for further
details.......
________________________________
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of Joe Shonk
Sent: 17 November 2010 16:28
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
I'm going to out on a limb and assume that you are referring to XenApp
and not XenServer.
First question, why Windows 2008 Standard x64?
Windows 2008 R2 is going to be a much better choice,
especially if you are running this directly on new hardware.
Second, Standard edition only support 32 gigs and you're
showing 36 gigs
Also, which metric is sitting at 25% utilized?
Joe
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On
Behalf Of James Scanlon
Sent: Wednesday, November 17, 2010 8:45 AM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: XenServer Slowdown
Greetings
(sorry about last unfinished email went to save to get more info and
clicked send - sigh)
We are having speed issues with our XenServers
* Brand New builds
* Deployed with Altiris
* Xenserver 5.0
* Windows 2008 Standard 64bit
* 36GB RAM
* Dual CPU, 8 Core processors X5560 2.80Ghz
* Appsense EM and PM (with citrix CPU and memory management
disabled)
* Sophos AV
* HP Hardware DL380 G6 with teamed nics
For normal operation we are getting 25 second login time (to desktop)
After approx 30 users (or 2 days without a reboot), the performance
drops to 1-1.5 minutes just to log in, everything sluggish, cant open
task manager even though multiple taskman icons appear in systray.
No performance issues on the systems themselves, all sitting around %25
utilised. (CPU, MEM, DISK etc)
So far:
We have disabled Sophos, no result
disabled Appsense, no result
turned on VT instructions for CPU in the BIOS, no result
turn off TCP offloading for the NICS
HW firmware for Servers / patching (latest PSP)
Software for OS and apps / patching
Any one seen anything like this already? apart now from bringing up a VM
host to eliminate hardware or a full manual rebuild and testing at each
stage of the build process - we are starting to run out of 'obvious
ideas'
ANY assistance and advice greatly appreciated!
JAMES
______________________________________
C. Hoare & Co. is authorised and regulated by the Financial Services
Authority with registration no. 122093. The FSA's address is 25, The
North Colonnade, Canary Wharf, London E14 5HS.
Registered in England no. 240822. Registered office 37 Fleet St, London,
EC4P 4DQ
Confidentiality Disclaimer:
This message and attachments are confidential and may be privileged, and
are sent for the personal attention of the addressee(s). If you are not
the intended addressee, any use, disclosure or copying of this document
is unauthorised. Information transmitted by email may be intercepted,
lost, destroyed, corrupted or delayed and as a result, C. Hoare & Co. do
not accept responsibility for any errors or omissions in the contents of
this message. If you would like to confirm the contents of this email,
please request a hard copy version.
If the contents of this message are of a personal nature, the email will
have been sent in a personal capacity and not on behalf of C. Hoare &
Co.
Monitoring/Viruses:
C. Hoare & Co. may monitor all incoming and outgoing emails in line with
current legislation. Although emails are screened for viruses, C. Hoare
& Co. cannot guarantee that any transmissions will be virus free.
________________________________________
Other related posts:
