Open guilhermesteinmuller opened 10 months ago
The workaround is inevitably ( when we are talking about octavia ) touching database.
First we need to locate the listener
select * from listener where id="ff617d44-65ca-4a06-80ee-26ffc9d0667c"
You should set the tls_certificate_id
column from the ghost secret container to NULL:
update listener set tls_certificate_id=NULL where id="listener_id";
We also need to delete the record inside sni
table:
DELETE FROM sni WHERE listener_id="listener_id";
Once thats done, you can remove the load balancer ( remember to use --cascade parameter )
This should probably be reported upstream to where ever the bug tracker du jour is these days for Octavia.
We have a case in which a user deleted a certificate that was set inside a listener, causing the lb being 'stuck' from being deleted.
From horizon or cli, we do not get any error message, just the 200 that the delete signal was sent.