Comment 13 for bug 1895723

i'm also running into this issue. I'm running a 3 controller 3 compute cluster, when keystone is restarted it gets stuck in a restart loop with the same error found in the docker logs above. The only way I get get keystone to start up is the change /etc/kolla/keystone/keystone-startup.sh token age to 8640000 (yes, 100 days because my deployment is older than Tomas' deployment). I'm running most recent version of master, so I picked up the keystone changes from about 28 days ago. I manually ran the farnet-rotate.sh script the cronjob runs and it seemed to complete ok. Each control node keystone and keystone_fernet container has 3 keys in /etc/keystone/farnet-keys folder so it appears the rsync is running ok.

```
(keystone-fernet)[root@ddc-control-3 /]# ls -ahl /etc/keystone/fernet-keys/
total 20K
drwxrwx--- 2 keystone keystone 4.0K Sep 25 12:32 .
drwxr-xr-x 1 keystone keystone 4.0K Sep 25 13:08 ..
-rw------- 1 keystone keystone 44 Sep 25 12:32 0
-rw------- 1 keystone keystone 44 Aug 27 14:23 1
-rw------- 1 keystone keystone 44 Aug 27 14:23 2

```

it just seems like keystone is using one of the older keys even know a newer one is present.

on a side note, when I pulled the updated git repo (from 350 to 380) I pip installed ./kolla and ./kolla-ansible then I ran kolla-ansible deploy to see if the keystone changes in the newer master 380 would fix the issue. has not for me. Not sure if I should be running deploy or upgrade so that could be part of the issue of why keystone changes did not seem to fix.