NIC dropping out randomly

[toadman]

After running 9.3 since it came out with essentially no issues, in the past week I've had my intel nic (82571EB - dual) drop out twice, seemingly randomly. By "drop out" I mean that it drops all connections and the switch shows it alive at 10mbps vs 1000mbps. Like it reset or something.

The server was running FreeNAS-9.3-STABLE-201511040813 when this started happening.

In my network config I run vlans on top of both em0 and em1 (and two other nics). The vlans on these two ports include the management vlan and one for cifs. So when the nic goes down of course nothing is reachable (which is how I discovered the issue). I have iscsi running on two other ports serving esxi boxes and they soldier on no problem.

To debug I just rolled back to FreeNAS-9.3-STABLE-201509282017 and moved the management vlan to a port on another nic. That way if it happens again at least I should be able to poke around on the server.

If this happens again, where is the best place to look for error messages related to the nic? And/or any other suggestions? Did any drivers get updated maybe? (Seems unlikely.)

Hardware:
AMD 555 x2
16 GB ECC
ASUS m4a89gtd
LSI 9211-8i
10 WD red, mix of 2TB and 3TB mirrors
Intel 82571EB (dual)
2x realtek 8111s

 

[BigDave]

If this happens again, where is the best place to look for error messages related to the nic?

# cat /var/log/messages
or if the logs are really long, try
# cat /var/log/messages | more

 

[anodos]

# cat /var/log/messages
or if the logs are really long, try
# cat /var/log/messages | more

Or
less /var/log/messages

Or "tail /var/log/messages | cowsay" :D

 

[toadman]

Ha!

Yes, tries /var/log/messages already. Nada. Will scan again if I can reproduce it.

 

[toadman]

Sigh, pretty sure this is a HW issue with the nic. Problem persists in multiple versions of FreeNAS from the last couple months.

 

[cyberjock]

Sigh, pretty sure this is a HW issue with the nic. Problem persists in multiple versions of FreeNAS from the last couple months.

Exactly. It's been well documented that Realteks just do "weird sh*t" when used, which is why we don't bother with them. We basically ignore any questions that involve "my network is slow and I'm using realtek" because we know better and have provided ample warnings.

Go get an Intel NIC from ebay for $10-15 and be happy. :)

 

[toadman]

Exactly. It's been well documented that Realteks just do "weird sh*t" when used, which is why we don't bother with them. We basically ignore any questions that involve "my network is slow and I'm using realtek" because we know better and have provided ample warnings.

Go get an Intel NIC from ebay for $10-15 and be happy. :)

I did get an Intel, and it went flaky. (You misread the post.) The Realtek's are saving my bacon in this case.

 

[cyberjock]

I did get an Intel, and it went flaky. (You misread the post.) The Realtek's are saving my bacon in this case.

Well, why the hell are you listening to me!

Just kidding.

Is it possible you have some Chinese knockoff? Could the card just be bad? Could the cable be bad?


The place to look for errors and such is /var/log/messages.

 

[toadman]

Ha!

We'll see. I moved that card to a different system (ESXi box) to test it further. Could very well be a card that went bad. I'll just let it sit and see if it drops out in that server.