HTTP resolver times out with a non 80/443 port

[bboure]

I have an http endpoint that listens on port 4000 and an AppSync HTTP Datasource that points to it.

endpoint: http://ec2-ww-xx-yy-zz.eu-west-1.compute.amazonaws.com:4000

When executing a GraphQl request, the resolver times out when executing the HTTP request.

From cloudwatch:

com.amazonaws.deepdish.http.HttpRequestException: io.netty.channel.ConnectTimeoutException: connection timed out: ec2-xx-yy-zzz-ww.eu-west-1.compute.amazonaws.com/xx.yy.zz.ww:4000",

Changing port to 80 works as expected.

[plus-]

I found the same thing.

the timeout URL was a concatenation of the hostname and a resolved IP:

"com.amazonaws.deepdish.http.HttpRequestException: io.netty.channel.ConnectTimeoutException: connection timed out: foo.bar.eu-west-3.elb.amazonaws.com/<THE_RESOLVED_IP_HERE>:4000"

switching back to port 80 solves it.

[linuxchata]

The same thing - HTTP endpoint datasource that is hosted port 5000 cannot be used by AppSync

[agustinnigrelli]

Has this issue any solution or workaround? i'm running in the same exact problem

[jpignata]

We will evaluate this change in a coming release. Please 👍 if this would help you.

[mstoyanovv]

Any progress on this? @jpignata

[cesardddp]

any news? still can't use a custom port?

[raducretu]

+1

[khaleeq-nicoin]

+1 why there are alot of limitations? you people also removed DATA API from serverless v2, and on HTTP resolvers port is now the limitation.

[nickathone]

+1 Given the documentation says this is possible, seems like quite a bad oversight. Very much needing to use non-80/443 ports for our platform setup.

And why has it taken over 3 years to be addressed?