Description

Sometimes customer may encounter routes that show the State "<Record Pending>":

 

"show route <prefix> detail" shows "State: <Record Pending>" for multiple routes when route-record is enabled even if there are no routing issues.

For route records, there is a function that marks a route head state to be Record Pending and adds it to the flash list. In some cases what happens is that there is no change in the Route head entry and thus it does not need to be flashed. Since it wasn't added to the flash list it never gets flashed and the state is never reset.

It is a time-based trigger. It checks when the route entry is made and when the last compare takes place. If a compare takes place before the entry, it would set the state to be record pending and send it to be added to the flash list. But before adding to the flash list it would check if there was any difference in the recorded entry and the new entry. If there was no change it would return it without adding to the flash list

 

user@hostname> show version
Hostname: XXXXXX
Model: ptx10001-36mr
Junos: 21.4R3-S1.6-EVO
Yocto: 2.2.1
Linux Kernel: 4.8.28-WR2.2.1_standard-gad6c183
JUNOS-EVO OS 64-bit [junos-evo-install-ptx-fixed-x86-64-21.4R3-S1.6-EVO]

user@hostname> show route X.X.X.X/15 detail
inet.0: 940708 destinations, 3889984 routes (937704 active, 0 holddown, 4183 hidden)
X.X.X.X/15 (3 entries, 1 announced)
State: <Record Pending>         <<------ HERE 


 

This can be seen on multiple devices but do not seem to cause any routing issues.

Symptoms

user@hostname> show route X.X.X.X/15 detail

inet.0: 940708 destinations, 3889984 routes (937704 active, 0 holddown, 4183 hidden)
X.X.X/15 (3 entries, 1 announced)
State: <Record Pending>         <<------ HERE 

user@hostname> show route active-path detail | match "Record Pending"

        State: <Record Pending>

        State: <Record Pending>

        State: <Record Pending>

        State: <Record Pending>

        State: <Record Pending>

        State: <Record Pending>

 

Trigger:

This issue might be seen if the following conditions are met:
* All Junos and Junos OS Evolved platforms
* Specific race condition is hit when "route-record" is enabled

 

Solution

JTAC observed a similar problem in JUNOS described  under PR - 1473746
The same issue is seen MX with high process memory and here in EVO we observed similar with msvcsd process. 

user@hostname> show system processes extensive no-forwarding
-------------------------------
node: re0
-------------------------------
top - 16:26:06 up 166 days,  4:51,  3 users,  load average: 3.54, 2.96, 3.48
Threads: 1932 total,   3 running, 1929 sleeping,   0 stopped,   0 zombie
%Cpu(s):  4.2 us,  2.2 sy,  0.0 ni, 92.9 id,  0.0 wa,  0.3 hi,  0.4 si,  0.0 st
KiB Mem : 64734472 total, 10144984 free, 22926588 used, 31662900 buff/cache
KiB Swap:  4194300 total,  4194300 free,        0 used. 39841540 avail Mem

PID USER       TID PRI  NI    SIZE     RSS S        TIME %MEM %CPU COMMAND
10636 root     27167  19   0  564728  508532 S 37-16:03:14  0.8 48.0 evo-cda-bt{WorkChannel-0}
21044 root     21827  19   0 1996632 2606236 R 33-02:57:00  4.0 36.0 msvcsd{msvcsd}   <---- HERE
27765 root     27995  19   0  904580  780840 S 25-01:34:17  1.2 28.0 packetio{fp-if-pool

In EVO check if a high msvcsd process is observed. Restarting that process may resolve the issue.

 

This issue is Resolved-In:

evo:22.4R3-EVO evo:23.2R1-S2-EVO evo:23.2R2-EVO evo:24.1R1-EVO junos:23.2R1-S2

Modification History

2024-01-04 : Article Created