fix-3238 gracefully shutdown server #3239
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 2 Ignored Deployments
|
|
@dillonstreator is attempting to deploy a commit to the Medplum Team on Vercel. A member of the Team first needs to authorize it. |
Apply Sweep Rules to your PR?
|
|
@dillonstreator - Sorry about the build issues... It looks like it's something funky with Coveralls. I have a fix here: #3246 |
| @@ -264,6 +265,7 @@ function addDefaults(config: MedplumServerConfig): MedplumServerConfig { | |||
| config.botLambdaLayerName = config.botLambdaLayerName || 'medplum-bot-layer'; | |||
| config.bcryptHashSalt = config.bcryptHashSalt || 10; | |||
| config.bullmq = { removeOnComplete: { count: 1 }, removeOnFail: { count: 1 }, ...config.bullmq }; | |||
| config.shutdownTimeoutMilliseconds = config.shutdownTimeoutMilliseconds ?? 30000; | |||
There was a problem hiding this comment.
This is redundant in its current state as the default timeout for http-graceful-shutdown is already 30 seconds https://github.com/sebhildebrandt/http-graceful-shutdown#options
I can remove, unless we'd like to adjust the default.
There was a problem hiding this comment.
30 seconds is appropriate. I don't have strong feelings about whether to be explicit or implicit here.
|
For my own curiosity, the AWS Fargate Task timeout is also 30 seconds: https://docs.aws.amazon.com/AmazonECS/latest/APIReference/API_StopTask.html In the common case, server shutdown should be fast (under 1 second). The exceptions being:
Those can take upwards of 30 seconds. Overall, it feels good to be explicit. Thanks @dillonstreator |
packages/server/src/index.ts
Outdated
| @@ -22,6 +23,17 @@ export async function main(configName: string): Promise<void> { | |||
| const server = app.listen(config.port); | |||
| server.keepAliveTimeout = config.keepAliveTimeout ?? 90000; | |||
| globalLogger.info('Server started', { port: config.port }); | |||
| gracefulShutdown(server, { | |||
| timeout: config.shutdownTimeoutMilliseconds, | |||
| development: process.env.NODE_ENV !== "production", | |||
There was a problem hiding this comment.
@codyebberson might be nice for the app config to define environment with an enum.
Maybe a separate issue?
|
@dillonstreator - Apologies for the inconvenience, but if you rebase & resubmit, the Github Actions problems should be fixed. Or let me know, and we can do it on your behalf. |
dillonstreator
force-pushed
the
fix-3238-server-graceful-shutdown
branch
from
4538d06
to
833067c
Compare
All good - I've rebased on |
|
Blerg, the prettier comment tool doesn't work on forks apparently, sorry about that. There's one prettier error in https://github.com/medplum/medplum/actions/runs/6778421837/job/18429916344?pr=3239 
|
chore(actions): add console output for `prettier-fmt` Actions workflow (#3278) feat(auth/external): make `PKCE` on external auth optional (#3279) server/fhir/repo execute independent io concurrently (#3227) MeasureReportDisplay (#3242) Create documentation on creating test data using batches (#3247) Clarify project admin vs. Super admin (#3267) feat(fhircast): add `FHIRcast` OAuth 2.0 scopes (#3274) Adding Health Gorilla setup documentation and samples (#3250) fix-3238 gracefully shutdown server (#3239)
* Release Version 2.1.12 chore(actions): add console output for `prettier-fmt` Actions workflow (#3278) feat(auth/external): make `PKCE` on external auth optional (#3279) server/fhir/repo execute independent io concurrently (#3227) MeasureReportDisplay (#3242) Create documentation on creating test data using batches (#3247) Clarify project admin vs. Super admin (#3267) feat(fhircast): add `FHIRcast` OAuth 2.0 scopes (#3274) Adding Health Gorilla setup documentation and samples (#3250) fix-3238 gracefully shutdown server (#3239) * Fix package-lock.json
Fixes #3238