This post is more than 5 years old
4 Posts
0
27248
Drops over 10G port and 1G Port
Hi,
We are using Dell powerconnect 6224 switch, sending data to its 10G port destined for 1G port (static mac binding) of around data rate of 700 - 800 Mbps but we packets drops continuously over both of the ports(got to know by snmp params of switch), and the 1G port is also connected to the receiver.
Please suggest something..
Thanks,
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
December 19th, 2012 07:00
To help ensure the switch is operating as smooth as it can, I would suggest running the newest firmware. 2.2.0.3 is pretty old version.
www.dell.com/.../powerconnect-6224
Here is a really good white page on flow control,
www.dell.com/.../Flow-Control-and-Network-performance-with-PowerConnect.pdF
With it enabled we may be seeing too many pause frames being sent, causing the degradation. We can look at these counters by running show command.
#show interfaces counters (specify the interface)
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
December 18th, 2012 08:00
What firmware revision is the 6224 at?
Is this a consistent packet loss, or sporadic? What percentage of loss?
What is at the other end of this connection?
What is the physical length of the connection?
Is this a new connection? or has this been working in the past and now is not? Any recent changes?
vippal
4 Posts
0
December 19th, 2012 07:00
Hi DELL-Daniel Co,
Thanks for your response..
firmware version: 2.2.0.3
packet loss are consistent, avg loss: 10%
Physical length is about 12-15 ft on both sides.
Its a new connection.
we have tried to disable flow control in switch settings, that stops the drops.
We also appreciate your suggestion over the below points:
Is Disabling this setting is the right way to solve the problem?
Is there any consequences of disabling the flow control if we are not overflooding the ports?
Thanks,
vippal
4 Posts
0
March 7th, 2013 06:00
Hi,
Running with same firmware and configuration, the drops shown by switch are the ifInDiscards(1.3.6.1.2.1.2.2.1.13), the references of this stats shows the reason could be the free-up buffer space..
My questions are:
1) Is Free-Up buffer space is only reasons for the ifInDiscards or there any other possible reasons ?
2) What is the buffer capacity and what amount of delay our application should make to get buffer free ?
Kindly suggest.
Thanks,
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
March 7th, 2013 09:00
1) Is Free-Up buffer space is only reasons for the ifInDiscards or there any other possible reasons ?
It very well could be other possible reasons, In order to get a better picture we need to look at the counters on the interface and see if we have a large amount of pause frames present. If you can run the command from above and post the output here.
2) What is the buffer capacity and what amount of delay our application should make to get buffer free ?
Here is the information I have on the buffer size for the switch.
6.1Mb per ASIC, shared across all ports. Single ASIC in 24 port models, dual ASIC in 48 port.
I however cannot find any information on delay times. And to my knowledge there is not a method to see buffer utilization.
Suggested steps.
1. Update the firmware on the switch.
2. Get the counters listed on the interface to see what level of pause frames we are seeing.
3. Knowing what type of traffic/application is being used on the network, and what NICs that traffic is being sent and received from.
4. Are Jumbo frames being used by any devices attaches to the network?
5. Are you currently using any traffic monitoring application such as wireshark?
Thanks