Interesting VRRP route tracking bug found in Junos version 15.1X49-D80.4 and below.
In our lab environment the vrrp route tracking was becoming disabled at what seemed random times.
To re-enable vrrp route tracking the quick option was to delete the tracking configuration from the Juniper SRX, commit, and then re-apply the vrrp tracking configuration, and commit.
So why was this happening?
After multiple failed attempts at trying to re-create this event, the last hell-mary idea came to mind, “have you tried turning it off an on again?”.
This may of seemed an odd idea at the time as we were not trying to fix something, we were trying to recreate a broken vrrp tracking event.
So why was this happening?
After multiple failed attempts at trying to re-create this event, the last hell-mary idea came to mind, “have you tried turning it off an on again?”.
This may of seemed an odd idea at the time as we were not trying to fix something, we were trying to recreate a broken vrrp tracking event.
To my shock after a successful reboot of the Juniper SRX the VRRP route tracking had become disabled.
It appears the odd default behaviour of vrrp route tracking in Junos version 15.1X49-D80.4 is to be disabled after a reboot.
We had found our random factor, power loss! Once an srx had lost power in the lab, and then is powered up (i.e. simulating a reboot), the tracking becomes disabled.
This surely could not be the desired effect Juniper wanted.
After speaking with Juniper JTAC it appears they had found this bug and have patched it in junos versions 15.1X49-D110 and above.
It appears they have fixed this bug, under the following public problem report:
PR1134189
“VRRP might get into unexpected state after GRES or RE switchover”
This did not match with our situation exactly, however after a bit more digging it looks like there is an associated internal juniper problem record:
PR1295987
“VRRP Route Tracking disable after chassis reboot”
PR1295987
“VRRP Route Tracking disable after chassis reboot”
Customer noticed that since rebooting one of the SRX345 in a VRRP cluster route tracking has not started again.
show vrrp detail
…
Virtual Mac: 00:00:5e:00:01:05
Tracking: disabled <<<<<<<<<<<<<<<<<< Disabled even if configuration says the feature is enabled
If the VRRP daemon is restarted, everything return to work.
Resolved In
14.1X53-D43 14.1X53-D121 14.1R9 14.2R8 15.1X49-D110 15.1R4 15.1F6 16.1R1
Junipers internal problem record.
A satisfying end to a fun bug hunt.
