cluster_reset has no effect on Debian

[eugenevd]

Hi,

On Ubuntu, my test case is:

1. define one db for postgresql_databases - run ansible = GOOD
2. add a 2nd db - run ansible = GOOD
3. remove either of them = NO CHANGE I then in vars.yml added postgresql_cluster_reset: true (since the default is false)

Looking at configure.yml I see for the drop and create cluster tasks the following condition: when: ansible_os_family == "Debian" and postgresql_cluster_reset and pgdata_dir_exist.changed

However, pgdata_dir_exists only changes when the data dir is dropped, which is limited to RedHat.

From my understanding pg_dropcluster + pg_createcluster is all that is needed in Ubuntu, as seems to be the configure.yml's intention, except for the limit; I reasoned I could solve the problem if for all the Debian steps I just remove that condition: when: ansible_os_family == "Debian" and postgresql_cluster_reset

This results in this error: unable to connect to database: FATAL: could not open relation mapping file "global/pg_filenode.map

Every second run fails, every run in between succeeds. (success, fail, success, fail..)

Am I getting this wrong, or is this a bug?

[aoyawale]

are you still having this issue?