Description

In this article, we explain a scenario where different behaviors are observed on a Virtual Chassis related to disruptions in the traffic with no clear indicators of a fundamental issue within the VC itself.

 

We highlight the types of logs that consistently appear and explain how the connection between a member acting as the master and a line card can introduce additional load on VC operations when errors are present. This behavior underscores the importance of monitoring VC‑port stability, as persistent errors not only generate recurring log events but also place stress on the control plane, leading to intermittent traffic disruptions.

Symptoms

  • VC intermittently dropping traffic to the VC, which is impacting SSH sessions and partially preventing SNMP polling. Reachability to the loopback and IRB interfaces is also affected.
  • Kernel and PFE logs indicate load on the VC with TCP keepalive timer expirations observed.
  • No signals of high CPU, Memory utilization or drops on the network interfaces.
  • VC‑port errors continuously increasing between the Linecard affected and the device acting as the Master. 

Solution

On this scenario we have a VC of 7 members, where FPC 1 is acting as the Master and is connected to FPC 7 acting as Linecard:

 

root@user> show virtual-chassis status

Preprovisioned Virtual Chassis
Virtual Chassis ID: xxxx.xxxx.xxxx
Virtual Chassis Mode: Enabled
                                                             Mstr           Mixed Route Neighbor List
Member ID  Status   Serial No    Alias-Name   Model          prio  Role      Mode  Mode ID  Interface
0 (FPC 0)  Prsnt    XXXXXXXXXXXX 0            ex4100-24p     129   Backup*      N  VC   1  vcp-255/1/0
                                                                                        2  vcp-255/1/1
1 (FPC 1)  Prsnt    XXXXXXXXXXXX 1            ex4100-48p     129   Master       N  VC   7  vcp-255/1/0
                                                                                        0  vcp-255/1/1
2 (FPC 2)  Prsnt    XXXXXXXXXXXX 2            ex4100-24p       0   Linecard     N  VC   3  vcp-255/1/2
                                                                                        0  vcp-255/1/3
3 (FPC 3)  Prsnt    XXXXXXXXXXXX 3            ex4100-48p       0   Linecard     N  VC   5  vcp-255/1/2
                                                                                        2  vcp-255/1/3
5 (FPC 5)  Prsnt    XXXXXXXXXXXX 5            ex4100-48p       0   Linecard     N  VC   3  vcp-255/1/2
                                                                                        6  vcp-255/1/3
6 (FPC 6)  Prsnt    XXXXXXXXXXXX 6            ex4100-48p       0   Linecard     N  VC   5  vcp-255/1/0
                                                                                        7  vcp-255/1/1
7 (FPC 7)  Prsnt    XXXXXXXXXXXX 7            ex4100-48p       0   Linecard     N  VC   1  vcp-255/1/0
                                                                                        6  vcp-255/1/1

 

root@user> show virtual-chassis vc-port 

fpc1:
--------------------------------------------------------------------------
Interface   Type              Trunk  Status       Speed        Neighbor
or                             ID                 (mbps)       ID  Interface
PIC / Port
1/0         Configured         -1    Up           25000        7   vcp-255/1/0
1/1         Configured         -1    Up           25000        0   vcp-255/1/0

--------------------------------------------------------------------------
Interface   Type              Trunk  Status       Speed        Neighbor
or                             ID                 (mbps)       ID  Interface
PIC / Port
1/0         Configured         -1    Up           25000        1   vcp-255/1/0
1/1         Configured         -1    Up           25000        6   vcp-255/1/1

 

Following messages are seen on FPC 7 and FPC 1:

 

Jun 23 07:57:12 user kernel: tcp_timer_keep: Dropping socket connection due to keepalive timer expiration, idle/intvl/cnt: 1000/1000/5
Jun 23 07:57:12 user kernel: tcp_timer_keep:Local(0x80000016:61958) Foreign(0x80000001:6157)
Jun 23 08:04:50 user dc-pfe[17457]: PFEMAN: Master socket closed
Jun 23 08:04:50 user dc-pfe[17457]: PFEMAN disconnected; PFEMAN socket closed abruptly
Jun 23 08:04:51 user dc-pfe[17457]: Routing engine PFEMAN reconnection succeeded after 1 tries
Jun 23 08:04:51 user dc-pfe[17457]: PFEMAN master RE reconnection made with 0x1000080
Jun 23 08:04:51 user dc-pfe[17457]: Probe exists for str bulkget_dfw
Jun 23 08:13:44 user dc-pfe[17457]: PFEMAN: Master socket closed

 

Each Packet Forwarding Engine (PFE) is connected to the primary Routing Engine (RE) and continuously sends TCP keepalives. When the kernel becomes busy, the PFE cannot process these keepalives, which leads to TCP timer expirations. As a result, the PFEs report TCP connection timeouts and disconnect their sockets, but they reconnect quickly once the kernel load subsides. In this case, the timeout occurred and corrected itself immediately.

 

If the Kernel remains busy for a longer period, it may continue dropping TCP keepalives, leaving TCP connections to all PFEs broken for an extended time. This condition can sometimes cause the Flexible PIC Concentrators (FPCs) on the device to reset and come back online.

 

No additional logs beyond those already observed were generated, and no daemon leaks were detected to explain the source of these messages and the inconsistent traffic behavior. However, when reviewing the VC‑ports, drops are incrementing on vcp‑255/1/0, which connects FPC7 directly to FPC1 (the master):

 

root@user> show virtual-chassis vc-port statistics extensive member 7 vcp-255/1/0

fpc7:

--------------------------------------------------------------------------
              RX           TX

Port: vcp-255/1/0
MAC control frames:    0           0
CRC alignment errors:   863740
Oversize packets:     0
Undersize packets:     0
Jabber packets:      529612
Fragments received:    18776

 

root@user> show virtual-chassis vc-port statistics extensive member 7 vcp-255/1/0

fpc7:
--------------------------------------------------------------------------
              RX           TX

Port: vcp-255/1/0   
CRC alignment errors:   1088581  > Increasing
Oversize packets:     0       
Undersize packets:     0       
Jabber packets:      616495    > Increasing
Fragments received:    24694   > Increasing

 

root@user> show virtual-chassis vc-port statistics extensive member 7 vcp-255/1/0    

fpc7:

--------------------------------------------------------------------------
              RX           TX
Port: vcp-255/1/0        
CRC alignment errors:   1091332  > Increasing 
Oversize packets:     0       
Undersize packets:     0       
Jabber packets:      617888    > Increasing 
Fragments received:    24775   > Increasing
 

root@user> show virtual-chassis vc-port statistics extensive member 7 vcp-255/1/0        

fpc7:
--------------------------------------------------------------------------
              RX           TX
Port: vcp-255/1/0         
CRC alignment errors:   1093677   > Increasing
Oversize packets:     0       
Undersize packets:     0       
Jabber packets:      619087    > Increasing
Fragments received:    24837   > Increasing


These VC‑port errors are typically associated with physical issues involving the cable, optic, or the port itself. They can significantly contribute to dropped control and data plane traffic between FPC 7 and FPC 1 (the master), resulting in instability across the entire VC. Since the errors are consistently increasing, the following troubleshooting steps are recommended:

 

  • Reset and clean the optic connected on PIC 1 port 0 bewteen the 2 members.
  • If errors continue to increase, replace the cables.
  • If the issue persists after replacing the cables, replace the optics and verify again.

 

After replacing the optic on vcp‑255/1/0 on FPC 1, the issue was resolved: log events stopped, and ping, SSH, and SNMP connectivity were recovered.

Modification History

2026-06-24 : Article Created