No ping-status 'OK' when script has no stderr/out?

[bdd]

No ping-status 'OK' when script has no stderr/out?

[bdd]

Looks like this is similar to #16. Django's development server did not support chunked encoding (header transfer-encoding: chunked and no content-length header). ...But you mention Apache, which should be implementing the standards rather properly. So that's a bit surprising.

When we thought we fixed chunked encoding in #16, we missed the fact that setting content-type: 0 is not necessarily enough for empty request bodies. Turns out since Go 1.8, the ambiguity of default zero value or explicitly set zero of content-length on HTTP request needs to be handled with http.NoBody or nil if the remote does not support chunked encoding.

Luckily this is essentially a single line addition. 047cb5f

I released v0.9.2 after verifying content-length: 0 header for empty bodies in status pings.

Can you try and report back?

[lfkdev]

It works now!

1.2.3.4 - - [21/Nov/2021:12:08:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/start HTTP/1.1" 200 5767 "-" "runitor/v0.9.2 (+https://bdd.fi/x/runitor)"
1.2.3.4 - - [21/Nov/2021:12:08:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/0 HTTP/1.1" 200 1125 "-" "runitor/v0.9.2 (+https://bdd.fi/x/runitor)"

I can see both requests in my hc instance with a script that has no output.
My setup is Apache/2.4.29 + mod_wsgi 4.9

Awesome, thank you!

[bdd]

I can't reproduce this with the healthchecks.io instance.

(Although not relevant to this, that PingStatus logging should not be in the output. It's just noise but I clearly made a git add -p mistake 🤦. Will publish 0.9.1 later today Fixed in 0.9.1.)

~% cat noout.sh
#!/usr/bin/env bash
  
foo() {
  bar=$(ls /)
}
foo

~% runitor -- ./noout.sh
2021/11/20 07:45:45 PingStatus(handle=<redacted> exitCode=0, body)

Can you verify from the HTTP server logs of your own instance, that in fact runitor does not make an HTTP request?

As explained in the release notes of v0.9.0, exit status is now always delivered. This commit slightly changed the URI paths hit by runitor.

v0.8.0:
start: <instance-url>/<uuid>/start
exit code == 0: <instance url>/<uuid>
exit code != 0: <instance_url>/<uuid>/fail

v0.9.0:
start: <instance-url>/<uuid>/start
exit: <instance_url>/<uuid>/<exit-code>

If the exit code == 0, Healthchecks interprets it as a successful run.

This feature was added to Healthchecks in v.1.18.0 (rel Oct 2020), with this commit.

Is your local instance version older or newer than 1.18.0?

[lfkdev]

My HC Version: v1.24.1

In both cases I can see requests in my apache log, here is the log entry with a script that has no output:

1.2.3.4 - - [20/Nov/2021:23:00:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/start HTTP/1.1" 200 5767 "-" "runitor/v0.9.0 (+https://bdd.fi/x/runitor)"
1.2.3.4 - - [20/Nov/2021:23:00:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/0 HTTP/1.1" 411 1090 "-" "runitor/v0.9.0 (+https://bdd.fi/x/runitor)"

and when it has output:

1.2.3.4 - - [20/Nov/2021:23:12:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/start HTTP/1.1" 200 5767 "-" "runitor/v0.9.0 (+https://bdd.fi/x/runitor)"
1.2.3.4 - - [20/Nov/2021:23:12:01 +0000] "POST /ping//xxxx-xxx-xxxx-xxxx/0 HTTP/1.1" 200 1125 "-" "runitor/v0.9.0 (+https://bdd.fi/x/runitor)"

The status code differs. 411 means "server refuses to accept the request without a defined Content-Length header", do you have an idea what could cause this?