Last Updated on 11 December 2021 by Lempod
Oct 23, 2021
Refreshed code and dependencies in Production.
Update the message for the time until the next engagement
- Old: You have to wait x hours before engaging a new post
- New: x hours until next post engagement
- Changed colour to be more visible in the new theme.
Reduced the number of unnecessary diagnostics which were causing the datastore to run slow.
Improved graph update reliability.
Fixed bug: Schedule graph updates from now, not based on a timestamp in the past.
Increased performance by improving the way that we count the number of remaining users to be engaged.
Oct 24, 2021
Tiered durations for graph updates if the post is still engaging (more frequent at first, less frequent as time goes on). This enables newer engagements to have priority, as requested by customers to ensure that their latest engagements are progressing.
Improved performance significantly by ensuring that we check the engagement’s state based on the new state parameter, instead of checking state by looking for the number of users remaining to be engaged.
Added tolerable delays under high loads, to ensure that all posts for all customers are delayed by only 10 seconds, instead of some customers receiving no post engagements at all.
Don’t update the graph whilst the post is in a draft state.
Oct 26, 2021
If a server is too busy, then the posts at the end of the queue for that server are assigned to another server.
Fixed bug: Some posts are not being marked as “stopped”, causing increased load on the datastore.
Added automated incident management functionality. If a post is delayed, then unassign it from the server and record the amount of time it was delayed (both the last instance and the cumulative delay time).
Oct 27, 2021
Large friendly text for paste LinkedIn link