[Bug 285751] [NEW PORT] sysutils/py-overlord: Deploy FreeBSD jails as fast as you code

From: <bugzilla-noreply_at_freebsd.org>
Date: Thu, 17 Apr 2025 18:57:39 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=285751

Jesús Daniel Colmenares Oviedo <DtxdF@disroot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #259328|0                           |1
        is obsolete|                            |
 Attachment #259644|                            |maintainer-approval+
              Flags|                            |

--- Comment #2 from Jesús Daniel Colmenares Oviedo <DtxdF@disroot.org> ---
Created attachment 259644
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=259644&action=edit
overlord-0.3.0.patch

Description:

* Update to 0.3.0

ChangeLog: https://github.com/DtxdF/overlord/releases/tag/v0.3.0

* Generate a random secret key by default:
  - By default, a secret key from the configuration file is used.
    However, if the user does not change it this will insecure his
    instance. Normally a user who knows what they are doing should
    change the secret key to generate tokens, however to be a little
    more secure by default, a random secret key is generated when the
    user does not specify one in the configuration file.
  - Remove metadata lock from memory when it is removed from the
    file system.
* Add cancel command (see
https://github.com/DtxdF/director/releases/tag/v0.13.0)
* Use aiostalk from a personal repo:
  - Due to a break change in greenstalk that breaks aiostalk
    (and also any application that imports it), I need to use the
    imported dependency from a personal repo in the meantime.
* Add Smart Timeouts:
  - Smart Timeouts disables a chain that fails until it comes back
    online.

    Technically, this does not disable the chain, but the chain does
    not appear in the API server's list of chains, so a client will
    only see (with a probability) chains that are known to work
    (until they don't), ensuring more stability than showing the
    client an unnecessary failure (and also reducing performance due
    to constant timeouts).

QA:

* portlint: OK (looks fine.)
* testport: OK (poudriere: 14.2-RELEASE, amd64, DOCS tested)

-- 
You are receiving this mail because:
You are the assignee for the bug.