Limit the compute job thread pool size

[garypen]

The router has always observed the APOLLO_ROUTER_NUM_CORES environment variable to restrict the size of the main tokio async job scheduler.

We are now enhancing the compute job thread pool to both respect this environment variable and restrict the number of threads in the pool.

If the environment variable is not set, then the size of the pool is computed as a fraction of the total number of cores that the router has determined are available.

If it is set, then the environment variable is taken as the number of available cores.

From this number, let's call it available, the router then uses the following table to size the compute job thread pool:

/// available: 1 pool size: 1
/// available: 2 pool size: 1
/// available: 3 pool size: 2
/// available: 4 pool size: 3
/// available: 5 pool size: 4
/// ...
/// available: 8 pool size: 7
/// available: 9 pool size: 7
/// ...
/// available: 16 pool size: 14
/// available: 17 pool size: 14
/// ...
/// available: 32 pool size: 28
/// etc...

This table should not be relied upon as an explicit interface, since it may change in the future, but is provided here for informational purposes.

[svc-apollo-docs]

✅ Docs preview has no changes

The preview was not built because there were no changes.

Build ID: a4d6d55042cffd8b0dab3381

[router-perf]

CI performance tests

• connectors-const - Connectors stress test that runs with a constant number of users
• const - Basic stress test that runs with a constant number of users
• demand-control-instrumented - A copy of the step test, but with demand control monitoring and metrics enabled
• demand-control-uninstrumented - A copy of the step test, but with demand control monitoring enabled
• enhanced-signature - Enhanced signature enabled
• events - Stress test for events with a lot of users and deduplication ENABLED
• events_big_cap_high_rate - Stress test for events with a lot of users, deduplication enabled and high rate event with a big queue capacity
• events_big_cap_high_rate_callback - Stress test for events with a lot of users, deduplication enabled and high rate event with a big queue capacity using callback mode
• events_callback - Stress test for events with a lot of users and deduplication ENABLED in callback mode
• events_without_dedup - Stress test for events with a lot of users and deduplication DISABLED
• events_without_dedup_callback - Stress test for events with a lot of users and deduplication DISABLED using callback mode
• extended-reference-mode - Extended reference mode enabled
• large-request - Stress test with a 1 MB request payload
• no-tracing - Basic stress test, no tracing
• reload - Reload test over a long period of time at a constant rate of users
• step-jemalloc-tuning - Clone of the basic stress test for jemalloc tuning
• step-local-metrics - Field stats that are generated from the router rather than FTV1
• step-with-prometheus - A copy of the step test with the Prometheus metrics exporter enabled
• step - Basic stress test that steps up the number of users over time
• xlarge-request - Stress test with 10 MB request payload
• xxlarge-request - Stress test with 100 MB request payload

[lrlna]

I am going to try backporting it to 1.x so I can run some more perf tests.

[lrlna]

@Mergifyio backport 1.x

[mergify]

backport 1.x

🟠 Waiting for conditions to match

• merged [📌 backport requirement]

[theJC]

Any chance you would be willing to incorporate the spirit of #6663 into this PR and log out the effective pool size and queue for diagnostics/info?

As a customer, I'd really appreciate having this available and be able to see the reality of these fairly important structures, based on the amount of cores dialed in for router ie, in kubernetes deployments.