This issue is fixed in VMware Tanzu for Redis 2.4. This bug will also be back ported to VMware Tanzu for Redis 2.2 and 2.3.
Workaround
Insert this workaround into the existing
/services/tls_ca procedure. Stop following the standard procedure when you reach the point in the
/services/tls_ca procedure that removes the old cert and then asks you to do the final
Apply Changes. Instead apply this workaround before proceeding to remove the old certificate.
Note: This procedure needs to be applied to all Reids On Demand Service instances with TLS enabled.
Download the service instance manifest
Replace
<GUID> with the actual GUID of your service instance.
bosh -d service-instance_<GUID> manifest > /tmp/redis-<GUID>.yml
Modify the manifest
Edit the manifest by moving the
update_mode key into the correct variable block.
Bad Example
- name: redis_certificate
type: certificate
options:
ca: /services/tls_ca
common_name: Redis Certificate
consumes:
alternative_name:
from: custom-redis-service-address
properties:
wildcard: "true"
common_name:
from: custom-redis-service-address
properties:
wildcard: "true"
features:
randomize_az_placement: true
update_mode: converge <<<<<<<<<<<<<<< BAD!!!!!!
Correct Example
- name: redis_certificate
type: certificate
update_mode: converge <<<<<<<<<<<<<<< FIXED!!!
options:
ca: /services/tls_ca
common_name: Redis Certificate
consumes:
alternative_name:
from: custom-redis-service-address
properties:
wildcard: "true"
common_name:
from: custom-redis-service-address
properties:
wildcard: "true"
features:
randomize_az_placement: true
Deploy the Redis On Demand Service Instance
Using the modified manifest, replace
<GUID> with the actual GUID of your service instance and start the deployment to trigger a leaf certificate rotation.
bosh -d service-instance_<GUID> deploy /tmp/redis-<GUID>.yml