Hi,

 

My system specs:

Intel Pentium G3258

16GB (4x4GB) DDR3 1600 at stock settings

Corsair CX400w (30A on +12v)

Hitachi 5k4000 4TB (Parity)

WD Green EZRX 3TB (data)

Seagate ST3000DM001 (data)

WD Green EARS 2TB (2) (data)

WD Green EZRX 4GB (data)

WD Black 500GB (cache)

Verbatim 2GB flash drive running Unraid 6.1.9 (had issues with 6.2.4, rolled back, was having same issues before 6.1.9l)

 

This setup has been running fine for over 18 months, and I clean the system out every 2-3 months. 

Hard drives are cooled by 2 x 120mm fans. 

CPU is cooled by a Coolermaster Evo 212+.

 

The issue I have been having for probably around 2 months is that I will randomly lose network connectivity to the system. 

-I cannot ping the Unraid IP, it has a static IP assigned

-I cannot ping or reach the Windows VM it is running.  Windows VM has a static IP assigned

-I cannot ssh or telnet to the Unraid IP

-I cannot connect to the http web UI

 

The things I have tried:

-connecting using a different LAN cable

-connecting to a different port on the switch

-connecting directly to the router

-installing a different network card (TP-Link Gigabit PCI-e)

 

There are 2 ways to resolve the issue temporarily

-reboot the UnRAID server

-unplug and replug the LAN cable.  This doesn't matter if it happens at the switch, router, or at the PC.  Disconnecting the LAN cable at any of these locations and plugging back in will provide network connectivity again. 

 

I haven't made any changes to my network. 

My other wired devices (android TV boxes, Linksys router acting as another access point, smart TV) do not lose connectivity. 

I've attached my diagnostic from when it failed this morning, I don't know at what point it did this. I hope there are some ideas. 

Let me know if more information is needed.

 

nas-diagnostics-20161207-0200.zip

Anyone with any thoughts?

 

I could try perhaps using a different Windows VM OS?  I see some have had issues with Windows VM's, though I have been running this Windows 8.1 OS for over a year now and this issue just started in the past couple months.

I could try a Windows 7 OS instead to see if that resolves the issue. 

* Rather unusual but the motherboard does not report any network chipsets.  I assume you have disabled it in the BIOS?

 

* You do have a Realtek network card, gigabit.  It drops the connection at least twice for a few seconds, but appears to reconnect.  Realtek has often been a problem, for many users.  If we don't find any other cause, I'd replace it, preferably with an Intel based one.

 

* Bridge name is xenbr0, so you configured this in the early days of the v6 betas.  The current bridge name is br0.  I really don't know if that's important.

 

* Possibly unrelated, but your VM is a monster CPU user!  And I mean HUGE!  It's possible that at its busiest, it's starving other subsystems for CPU time.  You may want to run 'top' at a command prompt now and then, and see how much it's using, and what the load averages are.  You may need a better CPU, for better response from that VM.

 

* Disk 5 is setup a little weird, it's drive capacity is not right.  It has a small HPA (an odd sized one I don't recognize), and it claims that the size of partition 1 is larger than the size of the drive!  As best I can tell, it corrects the numbers for you, so *might* be harmless.  If you ever have to pull the drive though, I'd run the hdat or hdparm command that corrects the sizing, and clear the drive so the partition table is correct.

 

* NTP isn't working because it's not set up right.  Go into the settings for time, and remove the word 'server' from each of the NTP server settings, just leave the domain (e.g. 0.ca.pool.ntp.org).

 

* The Realtek card is set up for jumbo frames with an MTU of 9200.  What is actually being used is 1500, everywhere.  I think it's best if the card is reconfigured to use 1500, not constantly be trying for 9200 (which would especially be difficult without matching network chipsets that can do 9200!).

 

Not sure if any of this is much help.  The networking appears to still be working at the end of this syslog, although it had just dropped and regained the connection.  There are multiple connections from 192.168.1.208.

* Rather unusual but the motherboard does not report any network chipsets.  I assume you have disabled it in the BIOS?

 

* You do have a Realtek network card, gigabit.  It drops the connection at least twice for a few seconds, but appears to reconnect.  Realtek has often been a problem, for many users.  If we don't find any other cause, I'd replace it, preferably with an Intel based one.

 

* Bridge name is xenbr0, so you configured this in the early days of the v6 betas.  The current bridge name is br0.  I really don't know if that's important.

 

* Possibly unrelated, but your VM is a monster CPU user!  And I mean HUGE!  It's possible that at its busiest, it's starving other subsystems for CPU time.  You may want to run 'top' at a command prompt now and then, and see how much it's using, and what the load averages are.  You may need a better CPU, for better response from that VM.

 

* Disk 5 is setup a little weird, it's drive capacity is not right.  It has a small HPA (an odd sized one I don't recognize), and it claims that the size of partition 1 is larger than the size of the drive!  As best I can tell, it corrects the numbers for you, so *might* be harmless.  If you ever have to pull the drive though, I'd run the hdat or hdparm command that corrects the sizing, and clear the drive so the partition table is correct.

 

* NTP isn't working because it's not set up right.  Go into the settings for time, and remove the word 'server' from each of the NTP server settings, just leave the domain (e.g. 0.ca.pool.ntp.org).

 

* The Realtek card is set up for jumbo frames with an MTU of 9200.  What is actually being used is 1500, everywhere.  I think it's best if the card is reconfigured to use 1500, not constantly be trying for 9200 (which would especially be difficult without matching network chipsets that can do 9200!).

 

Not sure if any of this is much help.  The networking appears to still be working at the end of this syslog, although it had just dropped and regained the connection.  There are multiple connections from 192.168.1.208.

 

-Yes, I have disabled the onboard as I am using the pci-e TP link card. I had thought perhaps the onboard was the issue but it appears to not be.

 

-Yes, this server was originally on 5.x.x but I have since upgraded and done clean installs, while maintaining the settings

 

-It's a windows 8 VM, so it's mostly just for downloading, shouldn't be using too much CPU but chrome can have a tendency to run amok at times. 

 

-Disk 5 was precleared before adding to the mix.  I was having this issue before I put this drive in to replace a smart failing drive.

 

-I'll configure NTP properly.  Thanks!

 

-I'll configure the Jumbo frames, I never touched this setting so it's odd that it is set so high.

 

-192.168.1.208 is my kodi box, I have it doing nightly synchronizations to my media to keep the library updated.

 

Thanks for any help.  I configured a Windows 7 VM, taking my Windows 8.1 VM down, and the whole system was online this morning so that's a positive (12+ hrs). I normally completely lose network connectivity overnight, or during the day, it's really at random and at random times.

 

One thing is that the unraid box completely loses network connectivity. At the console, I can't ping even the gateway, or any other devices on my network. Once I do the network cable unplug/replug, it works again.  Odd...

 

Thanks!