I'm seeing an odd occurence when using a cron-base...
# sstr
Ross Coundon
05/26/2021, 1:22 PMI'm seeing an odd occurence when using a cron-based lambda with SST.
When I fire up SST start, the target lambda gets hit multiple times, it almost seems to be that the SST somehow initiates a separate cron invocation in addition to the eventbridge rule/cron.
The rate is deliberately set quite low - (every 5 minutes) so invocations don't overlap when debugging.
Has anyone seen anything similar?
d
Dennis Dang
05/26/2021, 1:26 PMwhat's your cron rule set to? i thought yesterday a cron job was running on initialization, but turns out i had a bad rule.
r
Ross Coundon
05/26/2021, 1:26 PMIt's
Copy code
schedule: 'rate(3 minutes)',Ross Coundon
05/26/2021, 1:27 PM Copy code
const populatorCron = new sst.Cron(this, 'OmwPopulatorCron', {
schedule: 'rate(3 minutes)',
job: {
handler: 'src/main/handlers/populator.startLookingForMissingUrls',
environment,
timeout: 60,
},
});d
Dennis Dang
05/26/2021, 1:27 PMlooks correct to me. 😢
r
Ross Coundon
05/26/2021, 1:28 PMshame 😞
Ross Coundon
05/26/2021, 1:28 PMI'm going to try to remove the stack and debug on 1st deploy to see what happens
Ross Coundon
05/26/2021, 1:43 PMYeah, seems to happen only after starting, stopping and restarting
sst startf
Frank
05/26/2021, 5:04 PM
Hey @Ross Coundon, hmm.. my gut feeling is that when
sst startFrank
05/26/2021, 5:07 PM
When it happens again, can you check the CloudWatch logs for the cron lambda and see if some invocations are retries of previously failed invocations. They would have the same request id.
r
Ross Coundon
05/26/2021, 5:07 PMWill do, cheers
f
Frank
05/29/2021, 5:21 AM
@Ross Coundon I managed to reproduce the issue. It’s indeed retries from previously failed requests. I put in a fix in v0.23.0 to disable Lambda retries when deployed via
sst startr
Ross Coundon
05/29/2021, 7:09 AMThat's excellent, thanks Frank. Can you guys think of a way to extend the timeout for an API Gateway backed lambda beyond 30 seconds? I've been thinking about it myself but can't come up with anything, if that were possible, it'd be a real game changer for local development.
f
Frank
05/29/2021, 4:46 PM
Yeah, we might need to hack around this… something real hacky on top of my head is for the API to return a 301 redirect, and it will keep redirecting until the local Lambda respond… lol it’s kinda dumb
Frank
05/29/2021, 4:47 PM
I will an eye out for ways to get around the APIG limit
r
Ross Coundon
05/29/2021, 4:47 PMOne man's dumb is another man's genius