(MODULES-6875,MODULES-7487) - Fix mariadb mysql_user password idempotency#1195
(MODULES-6875,MODULES-7487) - Fix mariadb mysql_user password idempotency#1195eimlav merged 1 commit intopuppetlabs:masterfrom
Conversation
…ency The change introduced in mariadb 10.2.16 (and 10.3.8) where: > the `SET PASSWORD` command stores user passwords in the `mysql_native_password` plugin `authentication_string` instead of in the `password` column was addressed in puppetlabs#1135 This behaviour was short-lived. In 10.2.19 and 10.3.11 the old behaviour was restored. > The fix in the commit dd6e74c62a2 made sure that the password field is always set, because that was the original issue that broke third-party tools. See https://jira.mariadb.org/browse/MDEV-16774 This commit is based on https://tickets.puppetlabs.com/browse/MODULES-6875?focusedCommentId=645260&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-645260
|
Hi @alexjfisher . I have come across this same issue while adding RedHat 8 support for the module. I ran your fix through adhoc and it worked fine however when I ran against RedHat 8 (which is using MariaDB 10.3.10) the idempotency issues still occured. I added the COALESCE function suggested in MODULES-7487 and this appears to have fixed the failing tests. Would it therefore be possible to add the COALESCE function at https://github.com/puppetlabs/puppetlabs-mysql/pull/1195/files#diff-c6c504610358af96dcf6e3372321e142R21 ? Cheers! |
there should be either solution based on version pinning OR the COALESCE, but not the both to keep code somewhat clear |
|
@eimlav Is this with RHEL-8 beta? I'm wondering if they backported the fix into the 10.3.10 rpm?? The final release of RHEL 8 actually ships with 10.3.11, so maybe this isn't a problem after all? |
|
@alexjfisher Ah I see. I'll see if I can try it out on a released version of RHEL8. Thanks for the info! |
|
@alexjfisher Tested out your changes on GA RHEL8 and they passed. Nice work! |
The change introduced in mariadb 10.2.16 (and 10.3.8) where:
was addressed in #1135
This behaviour was short-lived. In 10.2.19 and 10.3.11 the old
behaviour was restored.
See https://jira.mariadb.org/browse/MDEV-16774
This commit is based on
https://tickets.puppetlabs.com/browse/MODULES-6875?focusedCommentId=645260&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-645260