Below logs appeared on PTX10008/PTX10016 or MX router after customer rebooted routing Engine 1. The article details information about the reason for below logs. 

May 29 04:36:23 2024 re0-labhost kernel: tcp_timer_keep: Dropping socket connection due to keepalive timer expiration, idle/intvl/cnt: 1000/1000/5

May 29 04:36:23 2024 re0-labhost kernel: tcp_timer_keep:Local(0x80000001:6234) Foreign(0x80000005:64662)

May 29 04:44:50 2024 re0-labhost kernel: tcp_timer_rexmt: Dropping socket connection due to retransmit timer expiration Local(0x80000004:6156) Foreign(0x80000005:64706)

May 29 04:44:51 2024 re0-labhost kernel: tcp_timer_rexmt: Dropping socket connection due to retransmit timer expiration Local(0x80000004:6171) Foreign(0x80000005:56077)

May 29 04:44:51 2024 re0-labhost kernel: tcp_timer_rexmt: Dropping socket connection due to retransmit timer expiration Local(0x80000004:7210) Foreign(0x80000005:63561)

May 29 04:44:53 2024 re0-labhost kernel: tcp_timer_rexmt: Dropping socket connection due to retransmit timer expiration Local(0x80000004:32006) Foreign(0x80000005:52028)



The reason for this tcp timer error is that TCP sockets will not try to retransmit data, but give up after some time and treat the connection is dead - and the associated socket closed. These Log messages are informational and are due to loss of communication towards other Routing Engine or FPCs.

Hints to identify the component that failed and resulted the logs

1] From the logs, we can see hex values like 0x80000001, 0x80000004 and 0x80000005. If you convert them to IP, you will get 128.0.0.1, 128.0.0.4 and 128.0.0.5.

2] Identify the Internal interface in your chassis connecting to Control board, other routing engine, or to the Host System. In this case, the interface is bme0.

3] Identify the address assigned to those interfaces. From the below output we can see that RE0 is assigned with 128.0.0.1/128.0.0.4 and RE1 is assigned with 128.0.0.5/128.0.0.6.  

{Master}

labroot@re0-labhost> show interfaces terse bme0 

Interface        Admin Link Proto  Local         Remote

bme0          up  up

bme0.0         up  up  inet   128.0.0.1/2   

                              128.0.0.4/2   

{Master}

labroot@re0-labhost> request routing-engine login other-routing-engine 

Last login: Tue Jun 18 04:08:24 from master

--- JUNOS 20.3X75-D36.12 Kernel 64-bit JNPR-11.0-20230201.7537bf2_buil

{Backup}

labroot@re1-labhost> 
 

{Backup}
labroot@re1-labhost> show interfaces terse bme0 

Interface        Admin Link Proto  Local         Remote

bme0          up  up

bme0.0         up  up  inet   128.0.0.5/2   

                              128.0.0.6/2  

4] Also, check chassisd logs in local or peer RE for any ECC, DRAM, DIMM issue like below. When such component becomes defective, it would make local RE having watchdog timer expiration and not able to respond to TCP flows. As the result, the RE lost of communication would happen.
$ cat chassisd | egrep -i ecc
Nov 23 21:46:25 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 24 01:15:02 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 24 01:29:07 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 27 14:43:35 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 27 15:16:24 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 27 16:21:28 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM
Nov 28 01:35:21 CHASSISD_SBE_DETECTED: Too many single bit ECC errors in routing engine DRAM

5] With the help of above outputs, we can now confirm that RE0 lost communication with RE1 and hence reported the "kernel: tcp_timer_rexmt: Dropping socket connection due to retransmit timer expiration Local" logs 

2025-12-04 : Minor correction, add a new scenario and category
2024-06-18 : Article Created