Allow Customization of waitUntilReady Timeout via Environment Variable

[ttsuru]

Description:

In the PHP-FPM runtime, when a request times out, PHP-FPM restarts. However, in some cases, the restart itself times out, leading to a state where the runtime cannot be started anymore.

By default, PHP-FPM is expected to start within 5 seconds. However, when using Symfony with multiple extensions and opcache.preload, this 5-second limit may not be sufficient.

I found that increasing the waitUntilReady timeout to 8 seconds resolved the issue.

Would it be possible to allow the waitUntilReady timeout to be customized via an environment variable? If a PR is needed, please let me know.

Thanks for the great project!

[mnapoli]

Wow 8 seconds is huge. But if that's necessary and it actually works, I think it's hard to argue against this suggestion.

👍 to add an environment variable to configure that.

[ttsuru]

I understand that AWS Lambda has a 10-second cold start limit, so the startup time must remain within this constraint.

I am using the ap-northeast-1 region, where the Init Duration was 2644.02 ms during a normal cold start. However, when a timeout occurred and PHP-FPM restarted, the restart attempt failed, taking 7074.87 ms.

I am unsure why PHP-FPM restart takes significantly longer than the initial startup.

XXX The PHP script timed out. Bref will now restart PHP-FPM to start from a clean slate and flush the PHP logs.
--
Timeouts can happen for example when trying to connect to a remote API or database, if this happens continuously check for those.
If you are using a RDS database, read this: https://bref.sh/docs/environment/database.html#accessing-the-internet
NOTICE: Terminating ...
NOTICE: exiting, bye-bye!
XXX Task timed out after 28.55 seconds
END RequestId: XXX
REPORT RequestId: XXX	Duration: 28547.96 ms	Billed Duration: 30645 ms	Memory Size: 512 MB	Max Memory Used: 216 MB	Init Duration: 2644.02 ms
LOGS	Name: cloudwatch_lambda_agent	State: Already subscribed	Types: [Platform]
--
Fatal error: RuntimeException: Error while starting PHP-FPM: Timeout while waiting for PHP-FPM socket at /tmp/.bref/php-fpm.sock in /var/task/vendor/bref/bref/src/FpmRuntime/Main.php:39
The function failed to start. AWS Lambda will restart the process, do not be surprised if you see the error message twice.
--
EXTENSION	Name: cloudwatch_lambda_agent	State: Ready	Events: [INVOKE, SHUTDOWN]
INIT_REPORT Init Duration: 7074.87 ms	Phase: invoke	Status: error	Error Type: Runtime.UnknownReason
START RequestId: XXX Version: $LATEST
Unknown application error occurredRuntime.UnknownReason
END RequestId: XXX
REPORT RequestId: XXX	Duration: 7592.93 ms	Billed Duration: 7593 ms	Memory Size: 512 MB	Max Memory Used: 184 MB
LOGS	Name: cloudwatch_lambda_agent	State: Already subscribed	Types: [Platform]
Fatal error: RuntimeException: Error while starting PHP-FPM: Timeout while waiting for PHP-FPM socket at /tmp/.bref/php-fpm.sock in /var/task/vendor/bref/bref/src/FpmRuntime/Main.php:39