Hello Maciej, thx for quick response.

Am 12.07.19 um 18:52 schrieb Maciej Delmanowski: >On Jul 12, dev wrote: >>since there is no forum, AFAIK, I just subscribed and hope get infos here >>instead of (mis)using github-issues… >There is no forum. I feel that creating one would add yet another place me and >other people would have to keep track of. Mailing list messages just come to >your usual mailbox, and you don't need to do anything to keep track of things. > >Although, an upgrade to Mailman 3 could probably improve the usability a bit, >with web-based forum-like system. But that would have to go through DebOps >first, so a new set of roles for Mailman, especially that mailman3 packages >are now available in Debian. I totally understand your point regarding the forum. since I have no clue about mailman3, within a forum it's just easy to search & find already answered questions and maybe complete configs as inspiration. why not drop the mailing list and replace it with a forum, where one can send mails to? just an idea…

>>Q: how to set lc_collate on postgres DB-creation? >Good question! At the moment the 'debops.postgresql_server' role uses the >'pg_createcluster' script to create new PostgreSQL clusters. There is an >option for setting up locale (via item.locale) parameter, but not specifically >collation. You can see the code here: > >https://github.com/debops/debops/blob/master/ansible/roles/debops.postgresql_server/tasks/manage_clusters.yml#L17-L21 > >Would setting the locale suffice in your case, or do you need to set collation >specifically? I suppose that adding the --lc-collate parameter with default >the same as --locale wouldn't hurt. I'm not sure how often that option has to >be differently set than the locale. > >Anyway, to change the locale for all clusters, you can set in the inventory: > >#v+ >postgresql_server__locale: 'en_US.UTF-8' >#v- > >For the default cluster only, at the moment the easiest way would be to >override the main cluster variable definition: > >#v+ >postgresql_server__cluster_main: > name: 'main' > port: '5432' > locale: 'en_US.UTF-8' >#v- in that case, it's not about different locales & lc-collate, it's more about different locales/lc-collates per database at the moment, I'm working on a customers server, which also uses atlassian software. atlassian suggests for jira a Locale=C: /Create a database for Jira to store issues in (e.g. //|*jiradb* |//) with Unicode collation./ |    CREATE DATABASE jiradb WITH ENCODING 'UNICODE' LC_COLLATE 'C' LC_CTYPE 'C' TEMPLATE template0;| but for confluence: it must not be 'C': /Next, create a database (for example //|confluence|//): / * /*Owner*// is your new database user (for example //|confluenceuser|//)/ * /*Character encoding*// must be set to //|utf8|// encoding./ * /*Collation*//must also be set to //|utf8|//. Other collations, such as "C", are known to cause issues with Confluence. // / since unicode collation is allowed with jira, I hope that it will work for my customer and will report back if not. other apps, like bitbucket, crowd … don't have any recommendations on locales & collations.