After adding apply-group configuration below the RPD core dumped. This lead to a service disrupting.

We need your help to investigate the logs and decode the core-dump.

set groups MAC-MOVE-LOG routing-instances <*> vlans <*> switch-options mac-move-limit 1

set groups MAC-MOVE-LOG routing-instances <*> vlans <*> switch-options mac-move-limit packet-action log

set routing-instances apply-groups MAC-MOVE-LOG

Feb 24 16:16:34 JTAC rpd[17875]: JTASK_ABORT: abort rpd[17875] version 24.2R2-S1-EVO built by evo-builder on 2025-05-30 08:23:38 UTC: Invalid argument, BACKTRACE: ^0x5590b0cea200 ^0x5590b0cb2f1e ^0x5590afd0e212 ^0x5590b0d19cd8 ^0x5590b0ce28bb ^0x5590b0cda87c ^0x5590b0d092af ^0x5590b0d0a42e ^0x5590ae62d77c ^0x7f91d21be5bb ^0x7f91d21be676

Feb 24 16:16:39 JTAC systemd[1]: routing.service: Main process exited, code=dumped, status=6/ABRT

root@JTAC> show system core-dumps

re0:

--------------------------------------------------------------------------

-rw-rw---- 1 root root 1566622210 Feb 24 16:17 /var/core/re0/rpd.re.re0.17875.2026_02_24.16_16_34_-03.tar.gz

total files: 1

As per the backtrace in core, it looks to be rpd reconfigure happened due to the groups addition.

We can see from config that they are having instances types virtual-switch & vrf, under routing-instances.

As per the group config, they were trying to add mac-move-limit under vlans. This will work fine for virtual-swtch instances, but can throw error for vrf type instances.

The rpd might not have accepted the config, leading to restart of rpd.

We tried adding the same CLIs in instance type vrf & got commit error as below.

root@JTAC# set routing-instances VRF-MGMT-POP vlans vl123 switch-options mac-move-limit 1

[edit]

root@JTAC# commit check 

[edit routing-instances VRF-MGMT-POP vlans]

 'vl123'

  'vlans' configuration can only be used for 'instance-type virtual-switch' or 'instance-type mac-vrf'

error: configuration check-out failed: (statements constraint check failed)

But this will go through when added as group. So this can lead to rpd inconsistency & cause rpd restart with core dump.

Workaround is to have the config added only in virtual-switch instances or add the vlan directly under instance instead of wildcards.

PR 1942540 was raised to fix this issue & below is the RCA from Engineering:

vlans object configured under wildcard group is getting created even when there is no foreground object configured where this wildcard groups is applied.

This is causing RPD to read that vlans object and crash. Ideally, RPD should not receive this 'vlans' object as it is marked as remove-empty and there are no children present.

The ifx is to throw commit check error when group config is applied on routing-instances globally, when it has unsupported instance types.