[DRIVERS-2710] Atlas cluster setup script should create a unique cluster per-invocation, not per commit Created: 30/Aug/23 Updated: 07/Sep/23 |
|
| Status: | Implementing |
| Project: | Drivers |
| Component/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Minor - P4 |
| Reporter: | Bailey Pearson | Assignee: | Bailey Pearson |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||||||||||||||||||||||||||||||||||
| Driver Changes: | Needed | ||||||||||||||||||||||||||||||||||||||||||||||||
| Downstream Changes Summary: | setup-atlas-cluster.sh and teardown-atlas-cluster.sh have been updated to ensure that they generate unique atlas clusters per task instead of per commit. both scripts now rely on the default evergreen expansion `task_id`. If a driver is calling these scripts while using the `include_expansions_in_env` or `add_expansions_to_env`, no changes should be necessary. If a driver is not providing the `task_id` to the setup and teardown script, the config must be updated to pass this expansion as an environment variable for the scripts. |
||||||||||||||||||||||||||||||||||||||||||||||||
| Engineering Lead: | |
||||||||||||||||||||||||||||||||||||||||||||||||
| Start date: | |||||||||||||||||||||||||||||||||||||||||||||||||
| Driver Compliance: |
|
||||||||||||||||||||||||||||||||||||||||||||||||
| Description |
Summarydrivers-evergreen-tools' setup-atlas-cluster.sh creates a unique cluster per-commit but not per invocation. This can lead to race conditions for drivers that have implemented both the search index management API and the FAAS failover testing where the CI tasks setup / teardown a cluster with the same name within the same CI run. We should add a unique identifier (a task id, a random number, ect) to the provisioned cluster's name so CI tasks do not interfere with each other. MotivationWho is the affected end user?drivers engineers. How does this affect the end user?Drivers may see flaky FAAS failover tests and search index tests. How likely is it that this problem or use case will occur?likely. If the problem does occur, what are the consequences and how severe are they?minor annoyance. Is this issue urgent?This ticket should be done soon so that as more drivers implement the new CI tasks, they are not adding flaky tests into their CI and so that drivers who currently have this can fix their flaky tests. Is this ticket required by a downstream team?no. Is this ticket only for tests?yes. Acceptance CriteriaUpdate setup-atlas-cluster to create a unique cluster name per-invocation instead of a unique name per-commit. |