## Version 0.11.0, released 2025-01-07

* The CI broker no longer crashes if it fails to create a trigger

  message from a CI event. Previously, if there was some problem, such

  as a repository getting deleted between the CI event being added to

  the event queue and it getting picked up for processing, the CI

  broker would crash. The idea was that if the trigger message can't

  be created, the event can't be processed in a useful manner, and so

  it's better to give up. What really happened is that systemd or

  another system would re-start the CI broker, which would pick up the

  same event, and again fail to create trigger message, resulting in

  another crash.

  Now the CI broker will log the failure, delete the problematic event

  from the event queue, and then continue with the next event. This

  makes the CI broker more robust.

* Some log messages have "extra data" in separate fields, instead of

  the free-form message field. The separate fields are typically text

  with values in the Rust debug format, so they're still not very

  easily consumed by other programs, but they're at least easier to

  see.

* Log messages related to a specific CI run now include the broker run

  ID as an extra field. This makes it easier to extract all the

  messages related to a specific CI run.

* Most log messages now have an extra `msg_id` field that contains the

  type of message, such as `AdapterConfig`, `BrokerRunStart`, or

  `CibEndSuccess`. The full list is in the `src/logger.rs` source

  file.

* The `cibtool log` subcommand can extract `cib` log messages from the

  systemd journal:

~~~

sudo journalctl -S today -u radicle-ci-broker -o json | cibtool log --log-level info

~~~

  The output of `cibtool log` is JSON, which is reasonably easy to

  post-process by another program, such as `jq` or a small Python

  script.