Slow SBS 2003 Performance

Microsoft Windows server 2003 for small...
May 13, 2010 at 09:28:46
Specs: Windows XP - Vista
I am running a Dell PowerEdge 2850 with 4
3.0GHZ Xeon CPU's, 6GB of Ram, 5 - 10k
drives setup RAID5 with SBS 2003. In the last
month or two the server has started running so
slow its causing delays in mail delivery,
network file access etc. Nothing has changed
on the network and we are current on our MS
Forefront Client Security Patches with no
detected virus's. The NIC is set to 1000Mbps
and plugged into a GB backplane. I have
checked perfmon and just see the usual
suspects when it comes to high utilization
processes (store.exe 850mb) etc but nothing
out of the ordinary. I have also noticed when I
stop and start the Exchange System
Attendant the server performance increases for
a short time then after an hour or so slows
back down. We only have 25 users all of which
are running Outlook 2007. I have checked just
about everything and the only thing I can think
it may be is excessive heat in the room the
racks located in. There is no AC in the room
and the servers do heat up quite a bit. Does
anyone have any ideas what could be causing
these speed issues?

See More: Slow SBS 2003 Performance

Report •

#1
May 13, 2010 at 09:53:31
"There is no AC in the room and the servers do heat up quite a bit"

That is a formula for hardware damage leading to failure.

When was the last time you defragged the drive?
Using the pae switch in the boot.ini?


Report •

#2
May 13, 2010 at 09:56:19
I am running a Dell PowerEdge 2850 with 4
3.0GHZ Xeon CPU's, 6GB of Ram, 5 - 10k
drives setup RAID5 with SBS 2003.

Well there's your problem right there.........Dell!

LOL

Just kidding.........6 GB's of RAM seems like not very much for a quad CPU server if you ask me. Depending on what you're running for software on that server and what roles it fulfills on the network, this may be your bottleneck.

If memory serves me, Exchange is a bit of a resource hog so if you have this server doing anything other than running exchange, lack of RAM could very well be your main issue.

If you do any online surfing whatsoever on this server (and I should hope it NEVER gets used as a desktop like that) then you may need to install and run antispyware software like malwarebytes.

Excessive heat won't slow a computer down, but it will make it reboot randomly if it gets too hot. Also, it's hard on a computer system and will cause components to break down sooner than they should.


Report •

#3
May 13, 2010 at 10:08:02
The room the servers sit in is just a small file room. Though it
doesnt have a direct AC vent it does pick up air from the
general office. I think they probably running a little hot, but
nothing to drastic based on what I have seen from other
servers.

Are you referring to an exchange database defrag or a basic
disk defrag. I have ran the analysis tool and it says the disks
do not need de-fragmented. As far as the exchange DB goes I
just started here a few months ago and I have to assume
there have been no debutil or isinteg run on them for quite
some time. I am planning to run it one of these weekends but
I cant believe that would make the file-print services and
overall speed be as slow as it is.

Will the pae work on SBS 2003? I thought it had a 4gb cap on
Windows 2003 except for Enterprise and Data center edition.
Even if I only get to use 4gb shouldnt that be enough for 25
people? There is usually no more than 15 or so on the server
at any given time.


Report •

Related Solutions

#4
May 13, 2010 at 10:27:53
Curt - Thanks. It's running Microsoft Forefront Client Security and
is patched up to date. I am planning on rolling a new server with
x64 Windows 2008 but I need to keep this one going for another
month or so. In 22 years of I.T I have never seen one slow down
this much while showing only 18% CPU and 2.5 of the 4GB's of
memory used. It literally cant even open task manager without a
3 minute delay.

Report •

#5
May 13, 2010 at 10:58:40
Disk defrag. You never ever believe the analysis tool. If you can't remember the last time you defragged that is a major problem for a server. Highly suggest you run it if not to just eliminate it as a source of your issue.

I would also suggest before running defrag you run chkdsk /f to confirm file system integrity.


Report •

#6
May 13, 2010 at 12:57:41
Try this,

http://www.msexchange.org/tutorials...

I like tip 3. Exchange eats memory like it was candy. Being that you only have 6 GB. Also, what other roles to you have enabled on this server? I tend to like Exchange on its own box doing nothing but Exchange.


Report •

#7
May 13, 2010 at 13:17:35
how about we check dns settings also since this can create huge network lag times. Please provide a post of an ipconfig /all from the server and a workstation.

I assume you have already reviewed the event viewer logs for errors?


Ace its SBS. You can't seperate Exchange from it.


Report •

#8
May 14, 2010 at 06:45:02
Thanks guys. Nothing in event log, its clean. The system
"thinks" its working as it should without any indication of high
CPU or memory utilization. My DNS resides on the SBS
server which is also the DC, BE 2010 Server, Web and SQL
Server. I did a AV scan which found nothing and I plan on
running a defrag on the exchange databases this weekend. I
tend to think store.exe is not releasing memory. I rebooted
the server this morning after applying a couple updates and
store.exe was only using 300mb or so and the server was
running PERFECT. Later in the day that process seems to
increase to about 800mb and thats when the server starts to
crawl.

Report •

#9
May 14, 2010 at 06:46:22
I was also wondering if maybe SQL was causing some of this? I
know we have our SQL DB's set to expand as needed and I
know our usage has increased. Are you supposed to purge any
transaction logs in SQL or anything of that nature?

Report •

#10
May 14, 2010 at 07:46:51
"which is also the DC, BE 2010 Server, Web and SQL
Server"

Well there's your problem. I can almost garantee that you are out of RAM.


Report •

#11
May 14, 2010 at 08:41:46
Thanks Ace. I've been thinking that as well. I just started here a
few months ago and inherited a developers idea of a well
designed server infrastructure. I plan on buying two new quad
core xeon servers with x64 2007 and just moving SQL, Web and
anything else I can to that box. The problem is I have to keep
this POS running until I get the other two deployed.

Report •

#12
May 14, 2010 at 09:12:36
SBS is DESIGNED to run all of this using only 4gig of ram. Only backup exec is the added server software.

A simple check on hard page faults would determine if ram was an issue.

SQL server should have maintenance plans in place that optimize the database and truncate the transaction log.

Trek1363
I don't see any mention of you performing the basics as I suggested of running chkdsk /f, defrag, post ipconfigs etc.

You always start with the basics and work up the ladder. Present mode is just random guessing. Can we get systematic on this?


Report •

#13
May 14, 2010 at 09:33:58
I ran Checkdisk and it didn't find anything nor did it indicate
the drive was fragmented. As far as the IP config I have listed
it below. I realize the server should be able to handle a
standard install-basic configuration of SQL, but the DB's we
use are huge and have grown by about 200 percent in the last
6 months or so. Is it safe to assume there would be such
thing as a DB thats too much for a SBS 4gb server to handle?
Earlier today I mentioned the server was rebooted then ran
perfectly? As of right now it has slowed back down to a crawl
and nothing has changed by way of server load since this
morning. Doing something as simple as ipconfig /all is taking
about 3 minutes to generate data.

C:\Documents and Settings\administrator.AVERY>ipconfig
/all

Windows IP Configuration

Host Name . . . . . . . . . . . . : server5
Primary Dns Suffix . . . . . . . : avery.local
Node Type . . . . . . . . . . . . : Hybrid
IP Routing Enabled. . . . . . . . : Yes
WINS Proxy Enabled. . . . . . . . : No
DNS Suffix Search List. . . . . . : avery.local

Ethernet adapter Server Local Area Connection:

Connection-specific DNS Suffix . :
Description . . . . . . . . . . . : Intel(R) PRO/1000 MT Network
Connection
Physical Address. . . . . . . . . : 00-14-22-20-81-6F
DHCP Enabled. . . . . . . . . . . : No
IP Address. . . . . . . . . . . . : 10.0.0.98
Subnet Mask . . . . . . . . . . . : 255.255.255.0
Default Gateway . . . . . . . . . : 10.0.0.1
DNS Servers . . . . . . . . . . . : 10.0.0.98
Primary WINS Server . . . . . . . : 10.0.0.98


Report •

#14
May 14, 2010 at 09:54:00
??? chkdsk never indicates anything about fragmentation. Good to hear chkdsk didn't find anything and your dns config is good.

A 200% increase in database size is significant but that is storage not processing. Please provide a post of chkdsk from the volume containing the sql databases for review.

Are you familiar with sql administration? We need to review the maintenance jobs.

And PLEASE run defrag! Thanks


Report •

#15
May 14, 2010 at 10:05:10
Not checkdisk, I ran the analyze tool from the same window as
defrag. As far as being familiar with SQL administration that is
my weak point. I dont have much experience as an SQL admin. I
will have to wait until tomorrow to run /defrag. It will kill the server
for sure based on what its doing now. The funny thing is the
server has moments where it stops lagging and runs great for
about 30 seconds, then you can watch it slow back down with
something as simple as hitting the Start button.

Report •

#16
May 14, 2010 at 10:06:45
Another thing I noticed is the drive lights are pretty much staying
active 100 percent of the time. Usually you can watch the drives
on an array blink with activity. These are staying on all the time.

Report •

#17
May 14, 2010 at 10:09:21
Nevermind, I had a brain cramp...running chkdsk now and I will get back to you with the results if the server doesn't crash ;)

Thanks!


Report •

#18
May 14, 2010 at 11:07:06
Here are the chkdsk results.

C:\Documents and Settings\administrator.AVERY>chkdsk
The type of the file system is NTFS.

WARNING! F parameter not specified.
Running CHKDSK in read-only mode.

CHKDSK is verifying files (stage 1 of 3)...
251328 file records processed.
File verification completed.
10 percent complete. (1 of 12735 large file records processed)
Deleting orphan file record segment 80651.
10 percent complete. (5001 of 12735 large file records
processed)
Deleting orphan file record segment 135212.
12735 large file records processed.

Errors found. CHKDSK cannot continue in read-only mode.


Report •

#19
May 14, 2010 at 11:20:28
Would it be unreasonable to assume that running chkdsk could
resolve the server problems? Right after the chkdsk ran and
deleted those two orphaned files the system almost immediately
began running normal! Did I pull a noob move and miss the
obvious here? Apparently your "start with the basic's" comment
was dead on. I will keep you posted but I think that did the trick.

Thanks a bunch!


Report •

#20
May 14, 2010 at 11:36:48
You have to run chkdsk /f to correct disk errors. VERY important you do so.

I need the final part of chkdsk that states the disk size/files etc


Report •

#21
May 14, 2010 at 12:56:33
This is what I got this time...

Microsoft Windows [Version 5.2.3790]
(C) Copyright 1985-2003 Microsoft Corp.

C:\Documents and Settings\administrator.AVERY>chkdsk
The type of the file system is NTFS.
The volume is in use by another process. Chkdsk
might report errors when no corruption is present.

WARNING! F parameter not specified.
Running CHKDSK in read-only mode.

CHKDSK is verifying files (stage 1 of 3)...
1 percent complete. (45133 of 251328 file records processed)
Deleting corrupt attribute record (128, "")
from file record segment 45333.
9 percent complete. (236196 of 251328 file records
processed)
Deleting corrupt attribute record (160, $I30)
from file record segment 240414.
Deleting corrupt attribute record (128, "")
from file record segment 240679.
9 percent complete. (241196 of 251328 file records
processed)
Deleting corrupt attribute record (128, "")
from file record segment 241515.
Deleting corrupt attribute record (128, "")
from file record segment 241516.
Deleting corrupt attribute record (128, "")
from file record segment 241552.
Deleting corrupt attribute record (128, "")
from file record segment 241554.
Deleting corrupt attribute record (128, "")
from file record segment 241556.
Deleting corrupt attribute record (128, "")
from file record segment 241557.
Deleting corrupt attribute record (128, "")
from file record segment 241582.
Deleted corrupt attribute list entry
with type code 128 in file 241704.
Deleting corrupt attribute record (128, "")
from file record segment 235689.
Deleting corrupt attribute record (128, "")
from file record segment 241689.
Deleting corrupt attribute record (128, "")
from file record segment 241705.
251328 file records processed.
File verification completed.
10 percent complete. (1 of 12743 large file records processed)
Deleting orphan file record segment 80651.
10 percent complete. (5001 of 12743 large file records
processed)
Deleting orphan file record segment 135212.
10 percent complete. (10001 of 12743 large file records
processed)
Deleting orphan file record segment 235689.
Deleting orphan file record segment 241689.
12743 large file records processed.
Deleting orphan file record segment 241705.

Errors found. CHKDSK cannot continue in read-only mode.

C:\Documents and Settings\administrator.AVERY>


Report •

#22
May 14, 2010 at 13:25:45
C:\Documents and Settings\administrator.AVERY>chkdsk

needs to be

C:\Documents and Settings\administrator.AVERY>chkdsk /F

Those are not good errors and indicate a much larger problem than anything previously discussed.

BUT BEFORE YOU PROCEED FURTHER get TWO tested as good backups.

This appears to be disk array failure or imminent disk array failure. It appears to me your heat issue may have effected cpu/memory/raid card.

Reboot and go into the raid card bios and look at the error log. Report back what you see.

I would HIGHLY advise you get a AC unit in that room. Room temp with no ventilation should not exceed 66-68 degrees. Understand that inside the server will range between 20 and 30 degrees higher. If that room is 69-70 it is way too hot for the server.

You have Backup Exec. Do you have the SQL module for backing up the sql databases?

Be prepared for the entire system to fail.


Report •

#23
May 17, 2010 at 07:32:45
I am looking at purchasing a portable AC unit for the room the
servers are housed in. I plan to run the chkdsk/f this weekend
and hopefully resolve some of the disk issues. While I was
here on Saturday I noticed the virtual memory settings for the
server had been set to half of what the system recommends. I
changed them to the recommended setting of 6172 and the
system still seems to be running significantly better than it
was before.

I have good backup data ( I run regular tests ). The drives are
configured RAID5 so if I don't lose more than one drive I
should still be fine.


Report •

#24
May 17, 2010 at 08:24:16
Did you defrag the drive before you set a larger pagefile?
You should have.
Is the pagefile static or dynamic?
You should set it as static [min and max values the same] but make sure the drive is defragged so you have that size of contigous space available or you will compound the problem.

You have disk corruption. This is not to be confused with a drive failure in a raid5 array. Two things can be responsible for this.
Physically memory and/or the raid card are periodically writing corrupt data to the drive. No raid level will protect you from this.
Second is where the OS processing is writing files to the volume incorrectly. This can be due to running processes or issues with the master file table.

I still need to see the final results of a chkdsk for the disk stats to review.

The reason for good tested backups is chkdsk /f can correct corrupt file entries but entire folders/programs can quit functioning/disappear due to this correction.

Very important! HOW ARE YOU BACKING UP YOUR DATABASES??????? BE via sql module?


Report •

#25
May 17, 2010 at 12:14:43
Yep, we have the SQL Module in BE 2010. I can kind of
understand your drive corruption theory, but why does the
system speed up to normal when you re-start the Exchange
System Attendant? The minute that store.exe process stops
taking 800mb of ram the system runs fine. I also noticed SQL is
using 1.6GB of memory today. I posted everything it showed
when I ran the chkdsk. Are you talking about the results of a /f?

Report •

#26
May 17, 2010 at 12:40:48
Glad to hear the databases are being backed up. You may wish to confirm that Master, msdb, tempdb are also backed up. Master is what tracks all the databases; its essencial.

When the system boots it reads files into memory.
If those files have holes in them...... this is why disk corruption is so important to prevent and address when found. You could be missing code in store.exe.

Here is an example of chkdsk from one of my 2003 servers concerning the part that is missing from your chkdsk post

35553103 KB total disk space.
28458052 KB in 68397 files.
23020 KB in 8457 indexes.
0 KB in bad sectors.
774663 KB in use by the system.
23040 KB occupied by the log file.
6297368 KB available on disk.

4096 bytes in each allocation unit.
8888275 total allocation units on disk.
1574342 allocation units available on disk.


Have you reviewed your event logs after a bootup for errors?

What do you mean you have to restart the system exchange attendant? Did you review the user account/password being used for this service?


Report •

#27
May 17, 2010 at 12:45:43
The process store.exe was chewing up quite a bit memory. I
know its supposed to release it as needed but when I restart
the System Attendant the system speeds up until the
process begins using 800mb of memory again. I checked my
event viewer and confirmed the nightly exchange maint is
being done and completing. After looking at that I am inclined
to think I have some disk corruption so I plan on coming in
tonight at midnight to run a chkdsk /f and a defrag. Should I
run the defrag before chkdsk? The defrag can take hours to
run and I have about an 8 hour window to get this done.

Report •

#28
May 17, 2010 at 12:49:08
Another thing I noticed is that the SQL process is using about
1.6gb of memory. I did some checking and the process seems
to be part of the Backup Exec SQL Database. The backups quit
running but the DB was still banging away. I killed that process
and it free'd up another 1.5gb of memory and after restarting the
SA it's still not bogging down yet.Should BE still be running a
huge DB file like that after the backups are done?

Report •

#29
May 17, 2010 at 15:48:55
We use the remote be sql agent to backup our servers to a central dedicated backup server. It only uses 3,848K

Our backup server running the backupexec.exe using 21,000K, beserver.exe uses 78,000k and bengine uses 27,000k

Might want to post what you are seeing in task manager

This can all be a distraction though. You keep looking at higher end functions without addressing the possible sources of the higher end function.

kinda like getting poor gas mileage and doing a full tuneup only to have no change in results because you forgot to check the tire pressure.


Report •

#30
May 18, 2010 at 07:02:35
Thank you. Near as I can tell the 1.5gb SQL process was
being used by an instance of Sharepoint that was no longer in
use. As I said before I just started here so I learn something
new every day. I ended that SQL process and have been
monitoring server performance which still seems to be doing
better than it was. To address the fragmentation issues I
purchased Diskeeper 2010 and have begun a defrag on the
servers drives. Right now the system is using 2.25GB of
memory with just under 9 percent CPU. I didnt get to run the
chkdsk /f last night but plan on doing it this weekend when I
have more free downtime.

Report •

Ask Question