cascaded `health` endpoint returns `null`

[jpmens]

I read in the source that

if this endpoint responds, the daemon is considered healthy

but wouldn't emiting something a bit more positive look a bit more healthy? ;-)

$ curl -sSf http://localhost:4539/health | jq .
null

Maybe a true for healthy, or even

{ "healthy" : true }

[tertsdiepraam]

That makes sense to me. Ultimately it should report more information about the system as well, i.e. how all the components of the system are doing.

[bortzmeyer]

The daemon says "Ok" which looks positive to me. Here is how I test it from Icinga (I follow the Nagios API):

#!/bin/sh

# Simple Nagios/Icinga plugin for Cascade. 

USAGE="Usage: $0"

TEMP=$(getopt -o "h" -- "$@")
if [ $? != 0 ]; then
    echo $USAGE
    exit 1
fi

eval set -- "$TEMP"

while true ; do
    case "$1" in
	-h) echo $USAGE; exit 3;;
	--) shift ; break ;;
	*) echo "Internal error!"; exit 3 ;;
    esac
done

logger "Test Cascade health"

E_OUT=$(/home/cascade/.cargo/bin/cascade health 2>/dev/null)

if [ $? -ne 0 ];
then
    echo "Cascade did not respond"
    exit 2
fi

L_CNT=$(echo ${E_OUT} | grep -E '^Ok$' 2>/dev/null | wc -l)
if [ ${L_CNT} -eq 0 ]; then
    echo "Cascade health failed with \"${E_OUT}\""
    exit 2
fi

echo "Cascade OK"
exit 0

[jpmens]

@bortzmeyer you are correct that the cascade health utility returns OK, but I used the REST API which returns null as described above.

[bortzmeyer]

@bortzmeyer you are correct that the cascade health utility returns OK, but I used the REST API which returns null as described above.

Ooops, yes, I missed that point, sorry.