All network activity stops after use of RDP in Windows 7 VMs

Just had this problem begin in the last few days.

 

I keep losing IP connection to my unraid server, I cannot ping it or ssh or connect to the GUI.

 

I check the LEDs on the port and they are fine, link up & flashing away, same at the switch end it connects to.

 

If I logon via the local keyboard and run an ifconfig xenbr0 the interface is up, runnning, broadcast etc with an IP address it all looks fine but if I run a ping out it just fails with destination unreachable.

 

The only way of recovering is to physically remove the network cable and reinsert, once forwarding is reported everything starts working again:

 

May 13 08:21:22 Tower kernel: r8169 0000:04:00.0 eth0: link down

May 13 08:21:22 Tower kernel: xenbr0: port 1(eth0) entered disabled state

May 13 08:21:51 Tower kernel: r8169 0000:04:00.0 eth0: link up

May 13 08:21:51 Tower kernel: xenbr0: port 1(eth0) entered listening state

May 13 08:21:51 Tower kernel: xenbr0: port 1(eth0) entered listening state

May 13 08:22:06 Tower kernel: xenbr0: port 1(eth0) entered learning state

May 13 08:22:21 Tower kernel: xenbr0: topology change detected, propagating

May 13 08:22:21 Tower kernel: xenbr0: port 1(eth0) entered forwarding state  <- at this point can ping again

 

I checked syslog but see no reports of any issues when the link goes down, is there any other file I can check?

 

I've changed to a different port on the switch to see if that helps and will also try replacing the cable but I'm doubting its a physical connection thing based on the above.

 

The switch was working for my other PCs which connect through it also.

 

Anyone had this before, got any ideas?

 

 

Regards.

1---  Try rebooting the switch(s) and router.  (Those devices usually have a embedded computer and restarting them will sometimes 'fix' a problem.)

 

2--- Change that cable.  (You want to eliminate all of the easy stuff before you go looking for the more exotic.)

 

3--- Look on the 'Dashboard' page and report if you see any errors under the 'Network' in the 'System Status' section.

 

4--- Next time you post, list your server hardware, include a full syslog, report any other things that you might have observed. (Example: Does it loose the connection while you are using it, or does it happen over night.  How often does it happen.  Any pattern to when it occurs...)

Some more testing.

 

A switch reboot or cable pull will get the link going again - note the switch works fine for my kit during these incidents

 

I've also tried a new cable and new port

 

And the big thing, this happens with another switch from a different manufacturer

 

No errors are reported on the dashboard or syslog when the issue occurs

 

 

I can pretty consistently reproduce this by RDP'ing into my VM, then play back a video via VLC - chances are during the video playback the fault occurs and I lose all access

 

I NEVER had this issue with Xen it should be noted (exactly same HW config), only since switching to KVM

I take it no one else has experienced this, or maybe not tried ?

 

I also tried with a different network card (USB Robotics one I think) and issue is still there and reproducible.

 

RDP'ing in, firing up IE and watching a youtube video also caused it.

You may want to try an Intel NIC.  It's surprising how many times network issues involving a Realtek NIC disappeared after using an Intel NIC.

Yes, I'll see if I can get hold of one, any specific model?

 

And if someone could test my scenario's above for me and show everything is fine that would be great 

 

But I used to do this all the time with exactly the same hardware under Xen and it worked flawlessly.

Yes, I'll see if I can get hold of one, any specific model?

 

 

This is the one that I used to replace my RealTek NIC when I had issues with audio dropouts and video strutting.

 

    http://www.amazon.com/gp/product/B001CY0P7G?psc=1&redirect=true&ref_=oh_aui_detailpage_o07_s00

 

Oh, you may have to shut off your on-board NIC using the BIOS.  Keep that in might as you are installing the new one.  However, you are using your server in a different way than I am so you might actually be able to use boht of them.

 

 

 

 

OK, just found one on ebay uk using Intel RC82540EM Chipset, will try that when it arrives.

 

If anyone with a similar config to mine (KVM, Windows 7 VM) could try RDP in then play video via VLC or IE and youtube playback for at least 10 mins. and confirm all is OK would appreciate it 

OK, just found one on ebay uk using Intel RC82540EM Chipset, will try that when it arrives.

 

If anyone with a similar config to mine (KVM, Windows 7 VM) could try RDP in then play video via VLC or IE and youtube playback for at least 10 mins. and confirm all is OK would appreciate it 

 

I don't have your same hardware, but have tested with multiple VMs and had no issues as you describe.

Thanks for testing, just hope the Intel NIC resolves the issue then.

So, disabled the onboard NIC in the BIOS and plugged in the new Intel NIC as recommended:-

 

Jun  8 14:13:00 Tower kernel: e1000 0000:03:00.0 eth0: (PCI:33MHz:32-bit) 00:07:                              e9:82:22:46

Jun  8 14:13:00 Tower kernel: e1000 0000:03:00.0 eth0: Intel® PRO/1000 Network                                Connection

Jun  8 14:13:03 Tower logger: /etc/rc.d/rc.inet1:  /sbin/ifconfig eth0 down

Jun  8 14:13:03 Tower logger: /etc/rc.d/rc.inet1:  /sbin/ifconfig eth0 0.0.0.0 p                              romisc up

Jun  8 14:13:03 Tower kernel: e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex,                                Flow Control: RX/TX

Jun  8 14:13:03 Tower kernel: device eth0 entered promiscuous mode

Jun  8 14:13:03 Tower logger: /etc/rc.d/rc.inet1:  /sbin/brctl addif br0 eth0

 

 

 

Situation is now worse, when I play video in the VM the whole system locks solid, all VMs and unRAID itself.

 

No response from local keyboard, the HDD LEDs all stop flashing, system just totally locks up, only recovery is power cycle 

 

 

I notice this now continually also:

 

Jun  8 14:41:50 Tower kernel: dmar: DMAR:[DMA Read] Request device [03:00.0] fault addr 100000000

Jun  8 14:41:50 Tower kernel: DMAR:[fault reason 06] PTE Read access is not set

Jun  8 14:41:54 Tower kernel: dmar: DRHD: handling fault status reg 3

Jun  8 14:41:54 Tower kernel: dmar: DMAR:[DMA Read] Request device [03:00.0] fault addr 100000000

Jun  8 14:41:54 Tower kernel: DMAR:[fault reason 06] PTE Read access is not set

Jun  8 14:41:54 Tower kernel: dmar: DRHD: handling fault status reg 3

Jun  8 14:41:54 Tower kernel: dmar: DMAR:[DMA Read] Request device [03:00.0] fault addr 100000000

Jun  8 14:41:54 Tower kernel: DMAR:[fault reason 06] PTE Read access is not set

 

lspci shows its the new NIC:

 

03:00.0 Ethernet controller: Intel Corporation 82540EM Gigabit Ethernet Controller (rev 02)

 

seems the network connection fails regardless of the VM now, so guess my PC does not like this card 

 

So, I upgraded a Windows 7 VM to RDP 8 and properly enabled it as per http://blogs.msdn.com/b/rds/archive/2013/04/09/get-the-best-rdp-8-0-experience-when-connecting-to-windows-7-what-you-need-to-know.aspx

 

The problem does not seem quite so severe now, last test I could RDP in and watch youtube for several minutes before I lost my complete network again.

 

I've knocked up a quick bash script that pings my default router and when this fails it runs ifconfig down then up on eth0 to recover.

 

Also upgraded to final V6 release, can supply diagnostics via PM if you like, let me know what else you need jonp - BTW, when you did your test were you using Windows 8 or 7?

I've tested both Windows 7 and 8.  Diagnostics would be helpful.

Don't seem to be able to able to add attachments to messages, is there a support email address I can send the diagnostics to?

Don't seem to be able to able to add attachments to messages, is there a support email address I can send the diagnostics to?

What problem are you having with attachments. Is it too large or something?

Attachments to messages not forum posts.

Attachments to messages not forum posts.

Attach to post. That's the usual method for help with diagnosing problems on this forum.

NO, don't fancy doing that so everyone can see, privacy et al.

 

They must have a support email, or just tell me what's needed and I'll run the commands.

NO, don't fancy doing that so everyone can see, privacy et al.

 

They must have a support email, or just tell me what's needed and I'll run the commands.

Limetech - About - Contact

NO, don't fancy doing that so everyone can see, privacy et al.

 

They must have a support email, or just tell me what's needed and I'll run the commands.

Just as a matter of interest it might be interesting to see what information (if any) in the diagnostics zip in your view would compromise privacy (all the files are text files).  The idea is to make sure that nothing confidential is included in the files included.

NO, don't fancy doing that so everyone can see, privacy et al.

 

They must have a support email, or just tell me what's needed and I'll run the commands.

Just as a matter of interest it might be interesting to see what information (if any) in the diagnostics zip in your view would compromise privacy (all the files are text files).  The idea is to make sure that nothing confidential is included in the files included.

 

There should not be any confidential information in the logs. Lime can only guarantee this for the base system (safe-mode). Add-on authors have to make sure that no confidential information is logged by their addons. The only confidential information that I've seen in log is e-mail and in a few cases e-mail passwords when debugging was accidentally left active. I don't think that you will receive any additional SPAM as a result of posting a syslog containing an e-mail address. This can be easily avoided by using a separate email account for the server.

 

 

Here's a list of things that the log may contain that are NOT useful to an attacker.

 

Flash GUID and key - Not confidential -- ask Lime.

Internal IP address.

MAC address. (your WIFI MAC addresses are in the clear. So anyone nearby can sniff them. If the attacker can physically attach to your LAN then this may help them.)

HDD serial numbers are not confidential. They cannot be used against you.

 

I'm certain that there are additional numbers that may seem confidential but they are not.

 

If you have a specific concern please let us know.

Just to add to what has already been said about diagnostics information.

 

A lot of the support for this product comes from other forum members, and the diagnostics were created with that in mind.

Don't want to get drawn into this just want my issue resolving. I'm not the Facebook generation. Syslog mover file names and share names one concern.

 

I'll continue in email to support.

Don't want to get drawn into this just want my issue resolving. I'm not the Facebook generation. Syslog mover file names and share names one concern.

 

I'll continue in email to support.

Could the diagnostic report be easily modified to omit lines that mover logs? I can imagine there might be some things there that some would consider private. Don't want to post filenames from PRON for example.