We just released the first release candidate of version 0.67.0 as an extraordinary release.
This release will not follow our usual release cycle that starts at every 20th of each month with the first release candidate and ends 7 days later with the final release because we are releasing a big change (can be a breaking change for who has custom builds) that will improve the server performance.
The main reason
The reason we are doing that is this Pull Request where we remove the cache layer and back to use the database for all queries. The PR contains a good explanation of the reasons behind the change and the expected results.
We have being working on that for more than 1 month and we used this changes at https://open.rocket.chat many times to check the results that were very good.
The release cycle
Since this will be out of our release process we don’t really have a date for the final release yet, we are targeting the day 20th of this month, but that can change.
The normal release at 20th/27th
This release will not affect the normal release, we will have the candidate of version 0.68.0 at 20th of this month and a final release at 27th which will include the cache removal as well, we could say that version 0.67.0 will be a more long release candidate period for a important part of version 0.68.0.
If you are running a custom fork you may have problems if you rely on the cache layer or
We removed the collection events related to the cache layer, example:
We removed collection methods
processQueryOptionsOnResult and all the code at file
We started removing the
room.usernames property a long time ago, but we was rebuilding it at the cache layer level, now we are removing it at all for rooms of type
p and ‘c’, the direct message rooms still have the property listing the 2 usernames involved.
Regressions and issues
Please refer to the original Pull Request if you have any problem related to this change, we will keep monitoring this PR to check for comments and linked issues.