Posted on May 26, 2021 4:32 am
 |  Asked by Ismail Kalolwala
 |  131 views
0
0
Print Friendly, PDF & Email

We are planning to upgrade the Arista 7160 Series switches to a newer code. We tried following from CVP:

 

  1. Checking MLAG Health – Successful
  2. Enter BGP – Failed

Reason – Interface Port-channel (Ethernet 31 and Ethernet 32) Traffic was not reduced as per the threshold in given time. Port-channel 2000 (Ethernet 31 and Ethernet 32) are the MLAG links which connects to Peer Switch.

Switches are on software version 4.24.5M and they are in production. I can see BGP entering in to Maintenance but not the interfaces.. couple of interface keeps on showing the traffic threshold violation.

 

Please advice

0
Answered on May 26, 2021 4:33 am

Hi Ismail,

Thanks for reaching out.

Maintenance mode Working: Maintenance mode(MM) checks a few features and states before the System enters into Maintenance mode. It checks for the interface rates. The System will enter MM only if the traffic rate falls below the configured threshold. MM by itself does not influence the rate of the traffic.

More information regarding BGP MM is detailed in the below link:

https://eos.arista.com/eos-4-22-0f/mlag-maintenancemode/

 

You can issue "show maintenance" command on the device to check the Aggregate Rate (Mbps) of In and Out traffic.In this scenario, Aggregate traffic might not be showing zero value and  traffic is continuously coming on the device on interfaces et31,32 and hence timeout would have happened.

If the traffic rates on the interfaces are not reduced below the default threshold of 100kbps then they would fail to enter BGP MM. The default traffic thresholds are set to 100Kbps but may need to be tuned/adjusted for different environments flow rates in regards to utilizing BGP maintenance mode for software upgrades.

switch(config-maintenance)# show maintenance profiles interface default
Interface Profile: Default
Rate Monitoring:
load-interval: 60 seconds
threshold (in/out): 100 Kbps
shutdown:
enabled: no
max-delay: 300 seconds

The traffic rate information when we enter the BGP MM is given in the below document :
https://eos.arista.com/toi/cvp-2019-1-0/bgp-maintenance-mode-and-mlag-issu-change-control-actions/#BGP_Maintenance_Mode

 

In this scenario we might need to make sure that the traffic rate is reduced below 100Kbps(default) on all the interfaces in order for the interfaces to enter BGP MM.

(or)

We can change rate-monitoring threshold value to a higher value than the amount of traffic received on the interfaces not entering MM.

For example if interface et31 is receiving around 450 Mbps of traffic (which is much higher than default 100kbps) then we need to set the rate-monitoring threshold value to 500Mbps in order for the interfaces to enter BGP MM.

Sample config:

switch(config)#maintenance

switch(config-maintenance)#profile interface test
switch(config-profile-intf-test)#rate-monitoring threshold ?
<1-4294967295> Threshold in kbps      >>convert Mbps to Kbps

switch(config-profile-intf-test)#rate-monitoring threshold 500000

Let us know if this helps.

 

Thanks,

Bhavana.

0
Posted by Ismail Kalolwala
Answered on May 26, 2021 3:14 pm

Hi Bhavana,

With ur advice, on a pair of switch which were in MLAG peering, BGP maintenance was successful. But this pair of switch does not have any host connected. Like you have already advice in your reply - we can set the traffic threshold to 450 Mbps , but will this impact the traffic , when we apply it on production switches which does have traffics on almost all links. Also what is the best way to find out BUM traffic - we suspect BUM traffic on the switch on which we tested BGP Maintenance.

Also, we have vF5 instances connected to some other switch, VIP's connections traffic is seen on the  Switch on BGP maintenance worked. Is this a BUM traffic... Can you please advice. Or navigate us to any document which shows how F5 works in vxlan environment.

 

Regards,

Ismail Kalolwala

Post your Answer

You must be logged in to post an answer.