Hello guys, I'm trying to track down a bug, which I don't think it's SST specific but generic lambda, and I was wondering if somebody has faced a similar issue and can shade some light. Long story short, we have a small percentage of outgoing HTTP requests taking a very long time to finish, in the range of 1 to 10 seconds, no matter the host. The lambda is inside a VPC with a NAT gateway. We have integrated with Datadog, and I cannot find a pattern for it, but something that looks weird is that every single trace, short or long, contains a dns.lookup and a tcp.connect call for every HTTP call, which makes me think that connection pooling is not working.

AWS_NODEJS_CONNECTION_REUSE_ENABLED

is on, so I'm kinda lost here. Any idea or suggestion on where to look at? Many thanks!

Hey @Franco Gotusso, are you getting this on

sst start

or

sst deploy

? Or both?

hey @Frank so far only with

sst deploy

, I haven't seen it with

sst start

but can give it a go if helpful