Re: SBS 2000 Possible memory leaks
- From: "Gregory Orton : SBS Admin" <ignoranceisbliss@xxxxxxxxx>
- Date: 23 Apr 2007 02:59:58 -0700
Thanks Susan,
I have set up a perf log to log every 30 seconds for the for-seeable
future monitoring:
MEMORY:
Non-Paged pool memory (which is what seems to be the resource running
out)
Paged-pool memory
PROCESSES:
Non-paged pool memory usage
Paged-pool memory usage
No of threads
Do you think this has a chance of capturing the offending process if
it happens again?
I know what you mean about it not being a "normal" SBS box - however
Dell Openmanage stuff is all hardware monitoring rather than anything
to do with the OS.
And these problems were appearing before Apache and mySQL were
installed - so i'm pretty sure it's not that.
Any other advice you have?
Gregory
***************************************************************************************************************************************************
On 22 Apr, 20:15, "Susan Bradley, CPA aka Ebitz - SBS Rocks [MVP]"
<sbrad...@xxxxxxxxxxx> wrote:
Gregory... that's not a normal SBS box if Apache and MySQL is on that
box, let alone the NOD32 and the Dell. :-)
I ran a SBS 2000 box for years with no issues of memory leaks. Set up a
perfmon snapping stuff at a certain time to capture this.
Gregory Orton : SBS Admin wrote:
Hi there,
I have an SBS network with 40 clients and running SBS 2000 on a Dell
PowerEdge 1800 Server (Dual Xeon's, 2GB RAM).
WE have the standard SBS programs on there minus MS SQL. However our
company intranet server (Apache , MySQL) also runs off of this system.
The only other software installed is NOD32 XMON server AV and Dell's
own Open Manage Server Admin stuff.
The problem here is memory leaks. Now I know all about Exchange's
built in design to use available RAM and dynamically give it all back,
blah blah, but with 2GB or RAM exchange is no longer an issue.
We get a memory leak every so often that completely cripples the
server. The VPN RPC cannot be reached, it's impossible to even log
into the console, exchange becomes unavailable and my event log is
fille with all sorts of exotic error messages from DCOM to exchange to
dll's and all sorts.
Whenever I get this issue I can tell when it started, because the
"System" log list an error from 'Srv' about the lack of non-paged pool
memory at the time it started to happen. The trouble is, we often
don't see this error til about 4 hours later. At which point, exchange
has already filled the application log every 0.5 seconds with its own
garbage, so I can never see the cause of the original error.
I'm pretty sure it's not exchange causing this leak, but I am also
pretty sure that the only way to diagnose this, is catching the server
"in the act", and using perfmon and poolmon to do this WHEN it
happens. Trouble is, it's a catch 22 situation. Once the server is
crippled by the memory leak, i can't start any monitoring programs.
This happens intermittently. It happened this morning, but it hasn't
happened for at least 5 months, but could happen tomorrow or next
week.
Please tell me that to diagnose this, real time monitoring is not the
only way? Or is there some way I could log all this and then make some
sense of what happened?
Greg
.
- Follow-Ups:
- Re: SBS 2000 Possible memory leaks
- From: "Jacky Luo [MSFT]"
- Re: SBS 2000 Possible memory leaks
- References:
- SBS 2000 Possible memory leaks
- From: Gregory Orton : SBS Admin
- Re: SBS 2000 Possible memory leaks
- From: Susan Bradley, CPA aka Ebitz - SBS Rocks [MVP]
- SBS 2000 Possible memory leaks
- Prev by Date: RE: New Problem SP1 can't find file specified
- Next by Date: RE: SBS and BCM - positioning ?
- Previous by thread: Re: SBS 2000 Possible memory leaks
- Next by thread: Re: SBS 2000 Possible memory leaks
- Index(es):
Relevant Pages
|
