v6.3.12 - loging into a file vs journalctl

[dulhaver]

I am setting up a 3-node etcd-6.3.12 cluster on 3 rhel9 vms.

currently I am stuck with the attempt to log into a file. I understand the default way is to log to journalctl.

I can not get etcd up with the the following ExecStart in my (--user) unit file

ExecStart=/opt/custom/etcd/bin/etcd --logger=zap --debug >> /opt/custom/etcd/log/etcd.log 2>&1

only reducing it to

ExecStart=/opt/custom/etcd/bin/etcd --logger=zap

makes the service startable

1. what am I doing wrong?
2. with the above working ExecStart I am logging to journalctl, correct? If that is the only way with 3.6 ... what is better with that then logging to a file?

A thought on the side ... would not logging to journalctl be a little 'unintuitive' when you think about running etcd inside a container?

[dulhaver]

[dulhaver]

Someone pointed out

--log-outputs 'default' Specify 'stdout' or 'stderr' to skip journald logging even when running under systemd, or list of comma separated output targets.

will try, whether a log file works as "output target" as per this definition.

[dulhaver]

getting etcd.service: Failed at step EXEC spawning /opt/custom/etcd/bin/etcd --logger=zap --log-output=/opt/custom/etcd/log/etcd.log: No such file or directory the log file definitely exists and is rw to the user running etcd.service though.

If I said --log-output=stdout,stderror, then what? as far as I understand stdout & stderror they are displayed to the terminal someone is running a command from ... as this happens in the background I would know, what to expect and how to &> this to the desired log file.