#74 healthcheck warning when referint-update-delay is not zero
Closed: Fixed 6 years ago Opened 6 years ago by ilias95.

As we want to stop supporting the async referential integrity update over time and obsolete the referint-update-delay attribute, we need a warning such as "we recommend you set delay to 0" when somebody is issuing the healthcheck command on an instance.


Metadata Update from @ilias95:
- Issue assigned to ilias95

6 years ago

Metadata Update from @ilias95:
- Custom field Review Status adjusted to review

6 years ago

ack from me, but with one change: the message can be improved I think. What about something like:

The referential integrity plugin has an asynchronous processing mode. This is controlled by
the update-delay flag.

When this value is 0, referential integrity plugin processes these changes inside of the operation that modified the entry - ie these are synchronous.

However, when this is > 0, these are performed asynchronously. This leads the only being able to 
have refint enabled on one master in MMR to prevent replication conflicts and loops. Additionally,
because these are performed in the background these updates may cause spurious update
delays to your server by batching changes rather than smaller updates during sync processing. 

We advise that you set this value to 0, and enable refint on all masters as it provides a more predictable behaviour.

Ack, but the patch doesn't apply to master. Do you mind rebasing?

Metadata Update from @firstyear:
- Custom field Origin adjusted to None
- Custom field Review Status adjusted to ack (was: review)

6 years ago

commit 3ad1f7f
To ssh://git@pagure.io/lib389.git
2a50b9d..709d572 master -> master

Metadata Update from @ilias95:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

6 years ago

Login to comment on this ticket.

Metadata