You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Now that we've got this project working for Ophan's Elasticsearch Cluster, I'm kind of excited by it's potential to automate a full cluster refresh rather than just 1 node at a time.
Thinking about the initial motivating use case for this project, I'd express it as:
"I don't want any nodes in my cluster to be more than 14 days old - because I need to have software patches for security updates applied within 14 days to satisfy GDPR"
At the moment we're achieving that with ENR by setting a cron schedule that ensures we rotate enough servers over the course of 14 days to never be out of date - but I'd like to propose that we change that by allowing the Step Function to refresh more than 1 box with each individual run. So when we run the Step Function and say "I don't want any nodes in my cluster to be more than 14 days old" it will respect that and loop round refreshing instances until it has satisfied that requirement.
Coming on to my new feature request requirement: fully refreshing a cluster (in the event of an urgent security update or just because a new artifact needs to be rolled out), I can now express that requirement as:
"I don't want any nodes in my cluster to be more than 0 days old"
...so the choice between gradually refreshing a cluster or doing the whole thing would come down to a 'maximumAllowedInstanceAge' parameter passed to the Step Function when it runs.
Now that we've got this project working for Ophan's Elasticsearch Cluster, I'm kind of excited by it's potential to automate a full cluster refresh rather than just 1 node at a time.
Thinking about the initial motivating use case for this project, I'd express it as:
At the moment we're achieving that with ENR by setting a cron schedule that ensures we rotate enough servers over the course of 14 days to never be out of date - but I'd like to propose that we change that by allowing the Step Function to refresh more than 1 box with each individual run. So when we run the Step Function and say "I don't want any nodes in my cluster to be more than 14 days old" it will respect that and loop round refreshing instances until it has satisfied that requirement.
Coming on to my new feature request requirement: fully refreshing a cluster (in the event of an urgent security update or just because a new artifact needs to be rolled out), I can now express that requirement as:
...so the choice between gradually refreshing a cluster or doing the whole thing would come down to a 'maximumAllowedInstanceAge' parameter passed to the Step Function when it runs.
cc @jacobwinch
The text was updated successfully, but these errors were encountered: