Server2008 Network card disconnecting every 5-10 minutes slow lan transfer speed

HI,

have a strange problem that im hoping someone can solve.

we have an SBS2008 Server that keeps dropping or resetting the lan network connection every 5-15 minutes.

Server: HPML350G6

NIC: Dual port Broadcom 10/100/1000

Problem.

the server was until this week in a very poor Broadband area (5.1kms from the exchange = 200k !)

so when RDP disconnected we always blamed it on the bandwidth as generally it would re-connect. however now it has a 100/100 PTP ->internet  fibre connection on the lan and its still doing it. even from within the lan.

the motherboard had a failure last week and was replaced under warranty, i was assuming that this would solve the problem with the network card as the drivers and firmware for the server were all up to date before it failed and it made no difference with the new board in.

symptoms are that if you attempt a long download it will time the download out  and disconnect RDP at the same time and occasionally cause dns problems.

im also getting a slow transfer speed over the LAN and WAN (600K/s on a gigabit lan end-end)

ive checked duplexing and manually set that im rapidly running out of ideas. anyone else had problems like this with server2008 ive noticed it on a pair of backend webservers as well running 2008
LVL 1
Wally_135Asked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

storkyIVCommented:
Have you replaced the RJ45 cables? Checked the patch panel / switch ports, maybe a dodgy connection?

Very strange though!
0
Wally_135Author Commented:
Yes cables replaced checked with an rj45 cable tester. switch verified as ok. we have multiple switches in that rack at the moment and its been tested with all of them.

any one had any success using wireshark to find broadcast storms and  packetflooding within a lan ? in case thats my Problem.

what gets me is that theres no obvious reason for it.
0
djaburgCommented:
Have you examined your system logs to see if there's any record of problems there?  Make sure there's nothing like power saving turning off the NIC to conserve power.  Is any other machine on the network having this kind of performance issues on the internet?  I encountered one client that had extremely fast service, yet their overall network speed was subpar.  It ended up being two things.  First was a MAC address of all 0's on the wan port of their router AND the auto-negotiation of connection speed would not work correctly, they too were on a fiber optic link.  The way around it was to manually set the speed of the connection in the router config and also to have the router copy the MAC of the computer we were using for testing.  I'd be more suspect of the auto-negotiation on the WAN connection though.  They were using a Linksys RV042 router.  
0
Webinar: What were the top threats in Q2 2018?

Every quarter, the WatchGuard Threat Lab releases an Internet Security Report that describes and analyzes the top threat trends impacting companies around the world. Are you ready to learn more about the top threats of Q2 2018? Register for our Sept. 26th webinar to learn more!

Wally_135Author Commented:
The nic has its duplexing manually set to gigabit and power saving turned off. but i dont think its the router. previously before they had the fibre link they were using a mid level draytek to route there internet over DSL and the disconnect problem was persistent then. however i will check the switchs asap as they were installed with the server they are HP curve gigabit switchs with a fibre module. ill check the negotiation on the router as well. just to make sure.
0
djaburgCommented:
Does any other computer have the issue of speed on the network, or is it exclusive the to server, or is it everyone and their speed to the internet?
0
aleghartCommented:
Sounds like a negotiation problem.  With 1000BASE-T, I don't think it will step down to 100BASE-T.  IIRC, GbE looks for all 4 pairs first so it can pass traffic bi-directionally.

100BASE-T uses separate pairs for send/receive.

Can you set the NIC to 100BASE-T full duplex?  If it doesn't drop, than I would re-examine the patch cable, anything punched down (jack or patch panels).

You may need more than a $10 LED blinker.  They only demonstrate continuity an proper pairing.  Will this wire carry 9VDC to the other end....it's not indicative that you can pass an Ethernet connection over it.  Just that you punched down the wires in the correct order.

Try something like a Fluke CableIQ or better for verifying or qualifying.  It's saved me a few times, knowing which pair is bad, and how far under the slab or into the wall the problem lies.
0
Wally_135Author Commented:
thanks for you input aleghart,

the patch cable is only going from the bottom of a rack to the switch at the top and the cable has been changed twice. and i made the second cable in case the factory patches were a bad batch.

im more concerned about the constant disconnection. The speed as you have all pointed out is likely to be negoatioation and therefore a different problem. and no other computers are suffering the disconnection problem except for DNS which is server managed.

0
aleghartCommented:
>bottom of a rack to the switch at the top

Strain relief boots and wire management loops the whole way?  Otherwise, a small move in the cable would exert significant force (relatively compared to the small-gauge wires) that could cause a problem with the physical connection.
0
Wally_135Author Commented:
Hi aleghart as i said the cable has been changed (at least 3 times) from manufactured cables to hand made ones.

what's wierd is the Nic keeps dropping out and its becoming urgent to fix it.

0
Wally_135Author Commented:
just to update. the system log doesnt report any errors only that services start again after the card has dropped out and reconnected. the HP management software doesnt report any problems with the card at all. only that the other port is diconnected (which it is).
0
DatedmanCommented:
Okay found this thread from the other one. :)

1) Just to make sure: any chance that you have clone the MAC address for this NIC to a device on the network?  (I know that sounds crazy but I have see a case like that here.)  
2) What services start again after the card starts up again?
3) Have you tried a different switch?
4) You talk about broadband and LAN both...is this server multihomed (more than one NIC?)  

Maybe you could post results of IPCONFIG/ALL (with a few things changed for paranoiac purposes [g].)



0
Wally_135Author Commented:
Hello all sorry been busy the last few days.

Ok.

tried today.

New port on switch
Different switch
Different Network card (Genuine HP compatible)
Different Cable

its made no difference, im coming to the conclusion that its something in windows server. ive already uninstalled the antivirus and stopped the firewall.

anyone had any experience with software corruption in Windows server 2008 SBS its definately dropping the whole network for a few MS.



0
DatedmanCommented:
I would try it with a vanilla network card like an old crappy 3-Com.  Also I would try this if possible: just one client on the network with it.  If that works, make it two, three, etc.

I am unconvinced that it is some generic Windows Server software glitch.  If so, seems we would have heard more about it by now!  I know I have certainly not experienced it altho I do not use SBS.  Still...IMO best way is to work with only one client then pump it until it breaks and figure out what made it break (assuming it works with one.)
0
Wally_135Author Commented:
ive tried it with the current domain members disconnected from the network. its still the same. 90% of the time when a download stops its at a random point in the download. but ive downloaded successfully a 1gb and a 512mb file and a 200mb at the same time. its driving me potty.

Datedman i was just wondering whether somehow a component has got damaged. ive literally changed everything hardware wise. and ive used another hp netowrk card (it bluescreened with the 3rd party card i had lying around)

i also ran this

Netsh int ip set chimney=disabled

to check whether it was dropping big packets which hasnt helped, im at  the point of re-installng the whole thing which would be a nightmare but might have to do.
0
DatedmanCommented:
Sounds like a driver problem.  I am not the guy for this but there are people who can use network monitor or whatever and figure out what the problem is I'm sure.  Have run into a couple at MS. If you open a trouble ticket with them it might take a little while to get to someone competent but likely you will get someone, and if you go through all the steps you've taken already with them you may get someone competent fairly quickly.
0
Wally_135Author Commented:
Hi is there anyone else that can possibly offer a solution here for this problem .ive never given up on a server before but it looks like i might return to sender on this one very soon. ive not been happy with the utter lemon HP have sold us. i have 350 g5's all over the place and they are perfectly behaved. this g6 has been a crippluing problem.
0
Wally_135Author Commented:
Ive solved this problem and unfortunately it was the most stupid solution ever!

WE use HP procurve switches but on this occasion we needed lots of POE so we went for linksys slm224p's (with an HP procruve 1gbe backbone).

Unfortunately the default IP of the 224p is 1.253 (same as the server!) which was the cause of our problem.

however when trouble shooting the problem we disconnected the server from the network to test whether or not there was a rogue IP on the same subnet. bu the 224p doesnt respond to ARP requests in the mode it was in and using a browser to see if there was anything that would respond  on the IP didnt work because we predominantly use FIrefox for our browsers and the config page for the 224p only seems to respond using Internet explorer ! something to do with using MSXML5.0 (which firefox doesnt have).

so remember that if you ever buy a linksys slm224 series switch ! it caused a major headache.




0
aleghartCommented:
Good find.

I manually assign IPs for everything but workstations.  Or, I'll use DHCP with static address.  Network gear will stay in one number range, servers in another, printers another, then a pool of dynamically assigned numbers for desktops, laptops, iPhones, etc.

I would have had the same problem if something came pre-assigned with a number on my LAN.

Easy fix...every LAN I make, I do _not_ use 192.168.1.x   or 192.168.0.x  or 192.168.168.x  as these are the most commonly used defaults for routers, switches, etc. (in that order).
0
Wally_135Author Commented:
yeah we used it as the network had half the subnet on a 1.X before and the sites so big i couldnt be bothered to change it all !

ive made policy now to keep all the switches assigned together.

thing was the server wasnt detecting a conflict at all. and wireshark didnt either. thats what threw me. there was no reporting. its was behaving like a dud NIC . damn you cisco.
0

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Broadband

From novice to tech pro — start learning today.