topic RE: HIGH RX PAUSE COUNT in ExtremeSwitching (EOS)

HIGH RX PAUSE COUNT

lino_calo_liber — Tue, 14 Mar 2017 18:02:00 GMT

I HAVE TYPED show port flowcontrol over my S8 core switch. This is the answer:
Here are some of the ports that have RX counts:

port TX Pause Count RX Pause Count
-------- --------------- --------------
ge.4.7 0 48751000
ge.4.26 0 87754431
ge.4.38 0 47367545
ge.5.34 0 28592459
ge.5.36 0 49402054
ge.5.38 0 158671480
ge.5.46 0 39917565
What does it mean? How can I clear the RX Pause Count in order to start counting from "zero" again. I would like to see the data increasing. Frequentely I lost the uplink connectio from S8 to distribution switch. Can I troubleshoot this issue?Thanks

RE: HIGH RX PAUSE COUNT

Daniel_Coughlin — Tue, 14 Mar 2017 18:15:00 GMT

RX pause is the counter keeping track of how many pause frames received on the port shown. What ever is attached to port ge.4.7 has sent 48751000. When trouble shooting this issue I simply run the command say 10 minutes apart and find the change. This will give you the rate that they are coming in at. The next step here is to look at the devices connected to the ports.

RE: HIGH RX PAUSE COUNT

Ronald_Dvorak — Tue, 14 Mar 2017 18:22:00 GMT

I'm surprised but it looks like that this basic function is not implemented ....

https://gtacknowledge.extremenetworks.com/articles/How_To/How-do-you-clear-port-counters-on-a-Matrix...

RE: HIGH RX PAUSE COUNT

Daniel_Coughlin — Tue, 14 Mar 2017 18:22:00 GMT

Internally the ability to clear counters has been debated many times. Having historical data is often helpful and spot checking for changes is not hard. On the S and K series engineering added the loop command which is great for just this kind of thing.

RE: HIGH RX PAUSE COUNT

Jason_Parker — Tue, 14 Mar 2017 22:53:00 GMT

Adding to Dan's comments
You could utilize the software EMC formally Netsight Console and use the "execute command" script tool (right click on the icon and select execute command script and type the following commands (for the ports in question) various times of the day (as it could be a factor of certain timed events) and then compare the files.

set length 0
show port flowcontrol ge.4.7
show port flowcontrol ge.4.26
show port flowcontrol ge.4.38
show port flowcontrol ge.5.34
show port flowcontrol ge.5.36
show port flowcontrol ge.5.38
show port flowcontrol ge.5.46

If you decide to check all of your switches with
set length 0
show port flowcontrol *.*.*

Then download a free tool called examdiff (it has a apple and pear picture) and compare the
files by show differences only between the any 2 collect

Makes life a bit easier
and also note the big picture
it could be a:
1. Mis-configuration of switch
2. Faulty NIC or wiring
3. Cabling
4. Low end PC,
5. Printers that should not be seeing traffic(like Multicast
6. Congestion
7. etc.

RE: HIGH RX PAUSE COUNT

Ronald_Dvorak — Tue, 14 Mar 2017 23:11:00 GMT

My opinion on the topic.....

If I troubleshoot I don't want to write down current counters and then 10 minutes later subtract that from the current counter to get the result for the change - I want to set them to 0 - during troubleshooting I've better things to do then solving math tasks or using some kind of app to show me the difference.

Another thing....there is nothing more annoying then sending a "show support" to the GTAC and then get the answer that some error counters have high numbers even that was because of some issue months ago and has nothing to do with the current problem that I'm working on.

For me "clear counter" is the most basic command that must exist on a CLI.

RE: HIGH RX PAUSE COUNT

Jeremy_Gibbs — Tue, 14 Mar 2017 23:11:00 GMT

I second this, the ability to not clear counters is absolutely ridiculous. You should at least have the abilities to set the option to clear the counters or maybe just lock out counter clearing if you so desire