Skip to main content
Start a Conversation

Unsolved

This post is more than 5 years old

Rygar1

1 Rookie

 • 

33 Posts

13213

November 8th, 2010 15:00

2816 + iscsi issues

Hey guys,

I had a 2724 switch in non manage mode connecting the file server (Windows 2008) using a Intel pro 1000/PT nic and a readynasNAS pro device using iscsi for our quick books data drive, everything worked ok till the switch went bad.  I bought a 2816 to replace the bad one and I did the following.

  1. Created a separate VLAN  ports 9-16
  2. Enabled Flow control on all ports
  3. Enabled Jumbo Frames on (set to 9012 highest allowed by the intel nic)
  4. Ingress Filtering on all ports
  5. Storm Control is off all ports
  6. Spanning tree is enabled under Global settings but its turned off for ports 9~16 (iscsi ports)
  7. Fast Link is unchecked on all ports
  8. ReadyNAS device has 2 NICs set to Round Robin Teaming

I been having issues were it runs really really slow and eventually just loosing the connection all together then after about 5~10 minutes everything is running fine again then at random times it will do it again.

  1. Should I turn off Spanning tree off all together? I assumed if I turned off on each port that iscsi is using would be ok?
  2. Do I need to turn on Fast Link for the iscsi connections?
  3. Do I need to set the speed on the network card and switch to 1Gb or Auto negotiation is fine? Speed is set to 1Gb
  4. Does Flow control need to be enabled or set to Auto negotiation?

 

Any ideas what might be causing the slow down and disconnection.

 

Thank you for your help.

 

 

 

bh1633

909 Posts

November 9th, 2010 06:00

Looks like your network is dropping packets.  With flow control enabled you should not be dropping any packets.

Is flow control enabled on the server and NAS NICS also?

What do the stats on the server and NAS show?:

On server and NAS Look for increasing errors in output of 'netstat -e'.   This indicates Ethernet errors that may be caused by bad cabling or a half duplex connection.

On server and NAS Look for increasing TCP retransmits in the output of 'netstat -s'.  If there are no increasing Ethernet errors (netstat -e) and there are retransmits, this indicates some network element is running out of resources and dropping packets.  If flow control is enabled, this should not happen.

On 2816, look at the RMON Statistics webpage and note any errors.  These are Ethernet statistics and errors indicate bad cabling or half duplex connections.

See if the NAS box has any other statistics that will give you a clue as to what is happening. (iSCSI stats).

Was this post helpful?

View All

0 events found

No Events found!

Top