plane
plane copied to clipboard
A distributed system for running WebSocket services at scale.
currently, spawns with ResourceLimits are untested. add tests for these. Additionally, it may be desirable to refactor the repeated nats etc invocations at the start of tests into a function,...
Clients should be able to connect to drones over IPv6, and DNS should serve AAAA records.
Currently, the spawn message bakes in assumptions related to Docker. We would like to support other execution backends down the road, so we should: - [ ] Move executor-related options...
In certain conditions (ie. when a nonexistent backend is getting polled very frequently), too many messages are logged to nats. Having a throttle would alleviate this issue.
This would introduce a struct to replace `grace_period_seconds`. I suggest: ``` struct ShutdownCondition { idle_threshold_seconds: Option, // replaces grace_period_seconds time_limit_seconds: Option, // total time limit } ``` Now that we...
This comes up a lot, and recently in #116. As a first pass, we can simply pass volume information on to Docker.
Since Spawner includes a Layer 7 proxy, it can refuse requests to a backend if they do not contain a bearer token as a cookie or as an `Authorization` header....
Two cases to check: - spawning a new backend - connecting to an existing backend Follow-up to #605
It's unclear based on the current documentation how spawn requests for already existing backends resolve conflicting grace periods. For example, if I spawn a backend for lock `foo` with a...
- [x] Make the executor configuration generic (#713) - [x] Turn executor config into an enum (#717) - [x] Make the executor itself generic - [x] Move pruning loop into...