PCs disapearing in a network

[joseLB]

I have a network with many PCs I will call "servers" (they are plain PCs (powerfull, 4-8 cpus, etc)). Some are win7 and others XP. Aleatorily some of them "disapear" or worst, just become "off" the newtork for a while.

This becomes a huge problem, as they have tons of scaned image files, running automatic processes on them.

On these "off" states, or the automatic process fails (long "offs"), or worst,in short "offs" they continues skipping some images.

My objective is to determine at the end what´s the cause of failure (network with cat-6, 3-com profesional switches, etc,).

In meanwhile, my idea would be a kind of ping on these servers. That´s a good ideia? If yes, what´s the best way to not have false pings or reduce them?

And, if anyone went thru this kind of problem, let me know, ok?

Thanks

Jose

[DicatoroftheUSA]

I think it is best practice to give anything that acts as a server a static address, it rules out most annoying small network issues.

then the usual, ping localhost > ping ip > ping name > check power settings, network adapters can be set to go to sleep ...

[joseLB]

I think it is best practice to give anything that acts as a server a static address, it rules out most annoying small network issues. then the usual, ping localhost > ping ip > ping name > check power settings, network adapters can be set to go to sleep ...

Thanks DicatoroftheUSA

But the PCs are in a local network, and each of them has a static IP. Good point on power setings, adapters, checked that and they where all "non-stop" (no sleep, hibernate, etc.). Seems the windows become lost in terms of network. Some suggested me to change all servers to w7, to not mix XP with w7...

[DicatoroftheUSA]

Well if it is an autoit script doing stuff, you can check the error codes of whatever function you are using, and have that trigger a function to test network connectivity.

A possibility is windows 7 limits concurrent connections to shares, I think to something like 20. I can't guess how many you using.

Edited July 19, 2012 by DicatoroftheUSA

[JLogan3o13]

Using the Ping function through AutoIt is fine, but if you have machines dropping off that often, you really need to do some investigation as to your network setup. The AutoIt script will only be a band-aid until you can figure out the root cause.

[gspino]

Several years ago, I wrote a script to monitor machines by ip or netbios name. Simply create a list, a simple text file where each machine name or ip address is on a separate line, and name this file 'servers.txt'. Put this list in the same location as the script and launch it. It will loop through this list and ping each machine name and a popup warning balloon will let you know if it's not pingable. I still use this every day. The source code is attached. (you can edit the servers.txt file on the fly)

This gives you a hand with troubleshooting, knowing when a machine goes offline, you can review the event logs during the timeframe it went offline. -Glen

ServerWatcher.au3

[gspino]

BTW, I've seen your issue before and found that the power management settings on the nic adaptor properties allows it to 'turn off this device to save power'. Remove that checkmark. See attached. -Glen

Edited August 7, 2012 by gspino

[Spiff59]

Last time I encountered something like that I found our Netgear 48-port gigabit switch had two fried fans and was cooking itself and suffering intermittant port failures. Turned out there were 3 months left of the 5-year warranty, and we upgraded from a hardware version 1 to a hardware version 4 switch for free. It fixed our issue, but I'm sure there could be a dozen reasons for what you're experiencing. Just wanted to contribute my 2 cents since no one had mentioned looking at your switch.