goss checks the name of the *binary*, not the name of the *process*. This can be checked from a user perspective via: | $ systemctl cat ngcp-rest-api | grep ExecStart | ExecStart=/usr/bin/node --title ngcp-rest-api /usr/share/ngcp-rest-api/server | $ sudo systemctl status ngcp-rest-api | grep PID | Main PID: 327024 (node) | $ ps -p 327024 -o comm | COMMAND | node Underneath, goss looks at the cmdline and stat files inside procfs: | $ cat /proc/327024/cmdline | ngcp-rest-api | $ cat /proc/327024/stat | 327024 (node) S 1 327024 327024 0 -1 4194560 38500 0 47 0 121 13 0 0 20 0 7 0 552217 704745472 32511 18446744073709551615 1 1 0 0 0 0 0 4096 82434 0 0 0 17 0 0 0 1 0 0 0 0 0 0 0 0 0 0 Since those two don't match with each other, our ngcp-rest-api service, running under the node binary, isn't identified as such a running process. NOTE: we're already checking for goss' service named ngcp-rest-api from within templates/510_init-daemons-ngcp.yaml.tt2, but this doesn't check if the process is alive. We need to explicitly check for the expected process name, which we're doing hereby. Change-Id: Iafa63b8e9fbfeee9edb2c0d48e8ed3514da4804amr10.3
parent
cf7a62fe2c
commit
d54de3e4da
Loading…
Reference in new issue