Junos running 18.1R3 experienced commit error and we issued in loading RSI and upon adding "set groups LOCAL-AS routing-instances <*> routing-options autonomous-system loops " to make the commit complete.
Commit error observed as below:
Core@JMX104# commit check
May 28 00:57:08
re0:
[edit routing-instances VRF-IN-SPOKE protocols bgp group VRF-IN]
'local-as'
Invalid loop count configured
error: configuration check-out failed
Following action plan we propose:
Once the working good configuration RE is decided, enter into config mode.
1.) Login to backup RE, take root user access
start shell user root
mgd -I <<< execute mgd -I, this will clear any database corruption seen on backup RE.
2.) Once step1 is complete, login to master RE.
Add following configuration
set groups LOCAL-AS routing-instances <*> routing-options autonomous-system loops 2
commit synchronize
3.) If step 2, does not work then take root user access on both the REs, and perform "mgd -I" on both the REs.
Master RE
mgd -I <<< execute mgd -I, this will clear any database corruption seen on RE.
Backup RE
4.) Once step3 is complete, Login to master RE.
5.) If step 4 does not work, login to backup RE, reboot the backup RE and perform commit on master RE
a.) request system reboot <<< on backup RE.
b.) On Master RE, commit following configuration while the backup RE is booting. This is to bypass config checks on backup RE.
commit <<< perform a commit.
c.) Once the commit is complete on master RE and backup is booted up, do commit synchronize on master RE.
commit synchronize <<< this should synchronize the working good configuration with the other RE.
See if step 5 resolves the issue.
If Step5 does not resolve the issue then proceed with step 6.
6.) Zeroize backup RE and commit the working configuration.
a.) Issue "request system zeroize local" on backup RE <<<< Care should be taken here, make sure 'local' is used, otherwise without 'local' it will zeroize both routing-engines.
The above command will completely wipe out the configuration and other files(present in var folder) from the RE where it is issued, so make sure to take a backup. This will ensure that the RE boots fresh without any inconsistent state.
b.) While the backup RE boots commit below configuration on master RE.
7.) If Step 6, does not resolve the issue, you should be ready with the good working configuration.
Then zeroize both the REs from master RE, the REs will boot fresh without any corruption, please be aware all the files stored under /var will be wiped out
>> request system zeroize <<< this will zeroize both the REs.
Load the configuration depending upon how the backup is taken
load set /var/tmp/<file-name>
load merge /var/tmp/<file-name>