Traffic problems when one VPC member is down

[from88]

Hello,

We've a "Classic VPC" scenario (VPC topology attached). We noticed - when one VPC member port is down (whole user port-channel in one NEXUS). We're starting to experiance issues with that user traffic. Sometimes it's very slow or with loss. We've tracked that somehow it can be related with VPC. Because when we disable the VPC functionality under the user interface (no vpc) the traffic problems dissapears and it's starting to work as it should even in "one legged" scenario. Seems like 'sometimes' it does some 'loop prevention mechanism' when really no loop exists.

Maybe you have some advice ? Thanks. 

our VPC options is here:

show vpc
Legend:
(*) - local vPC is down, forwarding via vPC peer-link

vPC domain id : 22 
Peer status : peer adjacency formed ok 
vPC keep-alive status : peer is alive 
Configuration consistency status : success 
Per-vlan consistency status : success 
Type-2 consistency status : success 
vPC role : primary 
Number of vPCs configured : 143 
Peer Gateway : Disabled
Dual-active excluded VLANs : -
Graceful Consistency Check : Enabled
Auto-recovery status : Disabled
Delay-restore status : Timer is off.(timeout = 150s)
Delay-restore SVI status : Timer is off.(timeout = 10s)

Software
  BIOS: version 07.51
  NXOS: version 7.0(3)I4(4)
  BIOS compile time:  02/15/2016
  NXOS image file is: bootflash:///nxos.7.0.3.I4.4.bin
  NXOS compile time:  10/24/2016 15:00:00 [10/24/2016 23:44:16]


Hardware
  cisco Nexus9000 C9372PX chassis 
  Intel(R) Core(TM) i3- CPU @ 2.50GHz with 16401544 kB of memory.
  Processor Board ID SAL2024RRSR

  Device name: JAYNET03A
  bootflash:   53298520 kB
Kernel uptime is 420 day(s), 13 hour(s), 20 minute(s), 51 second(s)

Last reset at 660524 usecs after  Thu Dec  1 00:06:53 2016

  Reason: Reset due to upgrade
  System version: 7.0(3)I4(3)

Seems like it'd drop traffic which is comming from CORE to NEXUS 5000 A (LEFT ONE) when through Peer-link to NEXUS 5000 B (RIGHT ONE) and sometimes it's gets dropped.

Maybe it's some bug ? Thanks

[marce1000]

 - Issue show logging , when this occurs; on the relevant switches  , client and peer (watchout for any errors); consider ugrading your platforms, to a later gold-starred release.

M.

[from88]

thank you, ofcourse logs is the first thing we check.

but... yes, maybe the next software would help, but it'd good to know if there's a bug or it's some "feature"..

[marce1000]

 - Yeah, not sure whether it's a bug either; going to a gold-starred release can be useful as a test in such cases.

M.

[Yachay]

Have you tried enabling peer-gateway?

[from88]

no, but i thought about that.

I can't understand how it would suit to existing problem.

As I know the Peergateway functionality is needed when your're using devices which doesn't support 'standard' ARP procedure like sotrage devices, some load balancers, etc.