Troubleshooting some strange server behavior

by Alex Kuretz on July 16, 2009 · 9 comments

in Windows Home Server

sad_mss

My primary MediaSmart Server EX475 had been not functioning correctly for a few weeks. This is the server I’ve had in production at home since late 2007, shortly before HP launched them for sale. It’s been backing up all the PCs in the house since that time, and I don’t really do any testing or add-in development on it in order to preserve it’s known good state. This is the box that I rely on to protect my data, stream my media, and allow me to access any files or PC while I’m away from home. Needless to say when it became unresponsive on the network, I had significant concerns.

The symptoms were entirely focused around network connectivity. The Windows Home Server Console would never connect, and the backups were no longer running at night. Remote Notification took a very long time to send email alerts. File transfer rates could be measured in the single-digit kilobytes per second at best and would frequently time out. Remote Desktop sessions would only connect sporadically, and I could barely stay connected long enough to check out the Event Viewer logs and determine that there were no reported errors. The health and drive LEDs were functioning normally, the server would shut down cleanly with a momentary press of the power button, and otherwise seemed to be working fine.

Every other device on my network was working fine, including my EX487 and development Windows Home Server. There was simply no explanation for the behavior I was seeing. The troublesome server was connected to my Netgear GS108 gigabit switch which lies at the center of my home network in the basement and had every port utilized. I had tried reconnecting the network cable, and even tried using a completely different network cable, both with no luck.

Finally out of desperation I began disconnecting all the cables from the switch, and connected the problematic server to a different Netgear gigabit switch. Instantly the server came alive on the network, performance was blazing fast and I was able to easily log in to the Server Console. I plugged the server back into the original port on the old switch, and it dropped off the network again. At this point I remembered that I had experienced an issue with this switch last summer while using it in a hot upstairs office, but since moving it to the cool basement I’d not had any troubles. I do find it pretty amazing that all the other ports on this switch were working fine the whole time, and just the one that my server was connected to experienced the failure.

I’ve since replaced that possessed switch, though I have it on my test bench to play with at a later time, or possibly take a hammer to it if I have a bad day. :) Have you had any strange hardware experiences like this, or what else would you have tried to troubleshoot the problem?





Article by

I'm Alex Kuretz, and I'm the founder of MediaSmartServer.net. I was the Lead Test and Integration Engineer at HP for the MediaSmart Server until April 2008 when I moved on to other opportunities outside HP. I've kept active in the Windows Home Server community, creating several add-ins and helping users make the most of their Home Servers.


{ 9 comments }

DamianP July 16, 2009 at 3:55 am

Interesting you mention this Alex. When I put my movies on my MSS I first rip them to my desktop PC, do a little jiggery pokery with them, and then transfer them over to my MSS (Both my MSS and my desktop PC are connected to the same DLink 8-Port Gig Switch). Normally my transfer rate would be anywhere between 25-35MB/s. Starting last week I noticed that the transfer rates dropped down to 10-12MB/s. Also, I was getting some stuttering on my streaming movies that I never got before. I started to troubleshoot a few days ago, but before I moved around ports on the switch I decided I would try to transfer files to my MSS from one of my HTPCs in my house that is not directly connected to the DLink Switch (if the switch was the problem then I should still experience the same issue). Of course, when I transfered files from my HTPC to the MSS my transfer rates were back to normal??? I then went back to my desktop PC and my transfer rates there were back to normal as well??? So I have no clue what happened, was it the switch, was it the MSS, was it something dare I say “supernatural” at work???

This has me thinking, I use the DLink DGS 5 and 8 port switches throughout my house (4 switches). For a little extra $$$ should I upgrade to a higher level switch like the HP ProCurve?

Phil Alban July 16, 2009 at 8:49 am

Hey Alex,

Part of my job is to maintain what was once an ad hoc network in a large public hospital. Much of what I was dealing with were older 100 MB switches. I have found that Linksys/Cisco switches often have the same trouble, with at least one port failing due to overheating, and the rest working fine.

It took me forever to trouble shoot down to the switch.

In fact, in some cases, fully utilizing all ports on a simple switch has actually “shut down” one of the ports. LOL, which meant that my solution was to always leave one port free. I feel your pain, though.

Thankfully, this hasn’t happened with my EX470, yet…

Greg July 17, 2009 at 1:51 pm

I have the 5 port version of that switch, and it too is flaky. I have it in a relatively hot 2nd story closet which does not get AC directly. It gets hot, and starts to reboot every few minutes.

If I let it cool overnight, it will be fine for a few hours, then do the same thing over again.

I have since replaced it with a procurve switch and have not had any issues.

Alex Kuretz July 17, 2009 at 4:05 pm

Damian, I can’t speak to the degraded performance issue, but I’ve seen some strange behaviors in a few different switches now. I think definitely keep them in a cool location if possible, and from what the other commenters are saying, maybe don’t fully utilize every port. You might consider utilizing a nicer switch as the central point in your network and maybe replace others if they die, but I wouldn’t go around throwing away switches unless you know there’s an issue.

Phil and Greg, thanks for sharing your experiences, it really does seem like the biggest killer of consumer electronic devices is heat.

Ryan July 24, 2009 at 12:07 am

Thanks for sharing this, sounds like heat is a killer. In my case, I think it was lightning. I have not had the issues you described with single ports flaking out within the home LAN, but my Netgear wireless G router’s WAN connection died a couple days ago and I think it was not coincidentally during a lightning storm.

I have a fixed wireless (radio) broadband ISP service, so I have an antenna on my roof with a CAT5 connection that goes through a small adapter then into the WAN port on my router. We had a close(?) lighting strike and my wife heard a ‘pop’ thinking it was the power surge protector, but all power stayed on. I guess the surge protector worked but the router port didn’t seem to. Now I’ve bought a Linksys wireless N gigabit, but I’m worried that it’s also susceptible to lighting with my setup.

Jeff J. August 6, 2009 at 8:15 am

I am having strange behavior with my server that is hard to explain. Let me try my best to explain. The strange behavior is similar to the Alex’s in the post above. At times my server works really well and at other times I can’t even connect to it. When I go to my laptop computer to check on my server, the internet is down and the network card says local only. There are a couple things that I did to fix it, but it’s only a temporary fix and it goes back to the same problem all over again. Normally, I would reboot my server or simply unplug the network cable and immediately, my network card goes from local only to local and internet. So for some strange reason I lose internet connection due to some issue. The other fix that I tried was to turn off my wireless card (using the on and off switch) on my laptop and then turn it back on and it will go from local to local and internet. This means that I don’t have internet connection and cannot connect to my server in-house or away from my home. I don’t know if there is a conflict between my wireless network card, if there is a conflict with my server or could this be a network issue. I had my server running well for a week now and I had to power it off because of a really bad storm. When I turned it back on it was working briefly, and then it went back down and my network card said local only when you put the cursor over the network that displays the wireless card. I am so frustrated that I decided to just turn off my server. I have a Linksys WRT320N Gigabit router (that I set-up and gave the server its own IP and reserved it) and MediaSmart Server EX470. I update the processor to Amd Athlon64 Ee Single-Core 1640 (2.7Ghz) Socket Am2 L2=512Kb and updated the memory to 2gigs. I contacted HP and they told me that they don’t have a troubleshoot department set up for the server at this time. If anyone of you can help, that would be greatly appreciated.

Scott August 10, 2009 at 9:29 am

The Netgear GS108 is a flaky piece of hardware. Doesn’t work well with some nics, (does strange things with my Linksys NSLU2, and flat out doesn’t work at gigabit speed with a realtek nic on a MSI motherboard of mine.) and many will start acting weird then die.

I’m on my third GS108, Netgear changed the warranty from 1 year to a life time.

Search the net on GS108 flashing lights or failure.

Comp1962 October 17, 2009 at 7:01 pm

I had a similar problem only it was my server causing the problem not a particular port on a network switch. The problem arose when I uninstalled hMailServer and it left behind some of its files and some other software components. Once those were removed my server and network have been playing well together.

There is nothing worse then software you install that does not properly remove itself and then gives a vague message that you need to do the rest.

I generally keep good notes on everything I do and in this case I was sloppy because it was late a night when I removed the software.

I do want to say one thing before anyone thinks I am implying hMailServer is bad. I am not. I just realized that to make it work properly you need to read quite abit of documentation before fully embarking on the task of setting up your own MailServer.

So the point here is make lots of notes and read as much as you can.

Paul Brown December 11, 2009 at 10:19 am

I’ve had EXACTLY the same problem recently!!! How weird – I also have an EX475 that I’ve had nearly a couple of years and plugged into the same 8 port gigabit Netgear switch. Recently I had connectivity issues (pings would time out more often than not) from my desktop machine. My desktop machine was also plugged into the same switch. However, when my desktop machine was not switched on, my other wireless based laptops would not have any problem connecting with the home server. As soon as the desktop machine switched on, the home server port would blink furiously. It seemed to be when the ethernet driver was loaded on the desktop machine and it turned into a gigabit connection. I changed the desktop machine hardware completely (needed an upgrade!) and installed Windows 7 but still the same issue – home server port blinking furiously when the desktop machine was running. So, I unplugged the home server from the switch and plugged it into the router (which is connected to the switch) and everything is ok. Admittedly, the home server is now plugged into a 100Mb port but it’s all good. So, I think the switch will go in the bin and a new one is required.

As a side note, the home server is on 24×7. The switch sits on top of the router and is extremely hot even though it’s well ventilated.

Comments are closed, visit the forums to continue the discussion.

Previous post:

Next post: