0

We have Dell R330 server with Broadcom NetXtreme Gigabit Ethernet cards, driver version 17.4.0.0 from Feb 2016, Windows Server 2012 R2 Std. x64 in a domain. The server is connected with a cable to a 3Com LAN switch - the same switch as other network servers, printers, workstations etc.

The server is normally available through tha LAN network but sometimes the connection gets broken for a fragment of second. The outage is long enough to be seen on RDP connection that shortly shows "Reconnecting" screen. Also sometimes a SQL Server connection fails from or to that server. It happens only occasionally, only at some times more often (e. g. sever times per hour). Even SQL Connection form the virtual Win7 PC on that server to SQL database on that server ocassionaly fails. Hyper-V, IIS, File services are installed.

In the event log there is only event 1058 The processing of Group Policy failed. every cca 2 hours but I do not think this is relevant.

The driver is updated but what can cause this, how to find out? No other network device behaves this way on our network.


I have run True Ping with delay of 500 ms and reallly the connection between the virtual host and the hosting server fails too.

4
  • 1
    Even SQL Connection form the virtual Win7 PC on that server to SQL database on that server ocassionaly fails - what virtualization is it? Server 2012 R2 with Hyper-V running Win 7, or something else? Because if that's failing, it doesn't sound like the physical network could be involved - if you unplugged the ethernet cable, they would still be able to connect, wouldn't they? [Edit: ah, you added Hyper-V] Mar 17, 2017 at 7:26
  • Did more testing & really even pings between hosting server and hosted virtual machine fail. Mar 17, 2017 at 13:35
  • 1
    What about other drivers and firmware? Dell R330 downloads - are they up to date? Does it have any software which might interfere with the network stack - any 3rd party firewall, any AntiVirus or Endpoint Protection installed? Is it regular through the day and overnight or limited to some specific times? Mar 17, 2017 at 18:41
  • A colleague changed some settings on the port on the switch and it helped, but did not share any knowledge about it so far... Mar 24, 2017 at 7:39

1 Answer 1

1

Go through the OSI layers:

  • start with physical, any errors on the interfaces CRCs etc... (you can also look into the switch if its managed as to what it sees)
  • does the interface drop in the switch logs?
  • if no "physical" issues: is there a port-channel/etherchanel/bonded or agregated interfaces if so any suspicious behaviour mac address wise...
  • how many interfaces does the server have connected to network
  • does the issue arise locally (same subnet) or just from afar. (routing , switching issue ?)
  • if you ping continuously to the server (at a fixed rate not a flood ping) do you loose packets if so whats is the error rate.

  • how is the port usage on server & switch side (packet drops? saturated bandwidth?) also check speed duplex on both sides.

Is the sql server going tru named pipes or does it go tru a switch?

6
  • There are no relevant errors on interfaces, only 3x Out of Recv. Buffer. Mar 17, 2017 at 7:37
  • is the switch managed or dummy switch ?
    – bob
    Mar 17, 2017 at 7:56
  • It is a managed switch. Mar 17, 2017 at 8:01
  • when the event happens does the interface in the log of the switch shows it as going down ? if not then its very likely the server has a software issue (it depends on how its all setup but running IIS , windows, hyper-V a SQL database and a win7pc on the server is quite a taxing thing) edit: also how many servers are on the switch ?
    – bob
    Mar 17, 2017 at 8:09
  • The port looks ok on the switch at the times of failure. Mar 18, 2017 at 6:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .