Closed Camelron closed 2 weeks ago
Should we also do the same with all of the other IRQBALANCE_* variables in irqbalance.env?
I guess it wouldn't hurt, but its really making me question the validity of what systemd is doing here. There are many cases in which environment variables are optional - most code in fact does the check:
var = getnev("ENVVAR");
if (!var)
var = <default value>
So I really don't see how systemd can consider it an error. Informational, or a warning perhaps, but certainly not an error.
I've opened https://github.com/systemd/systemd-stable/issues/425 to enquire about this
Systemd now logs when a variable expanded in a unit file is not explicitly defined. Thus, in the systemd service when we expand the variable a warning is logged https://github.com/Irqbalance/irqbalance/blob/master/misc/irqbalance.service#L11. This behavior was introduced to systemd in versions 254+ https://github.com/systemd/systemd-stable/commit/f331434d13488425ccd8485327085d15f8f92aea
Squelch this warning by uncommenting IRQBALANCE_ARGS in irqbalance.env, instead setting it to the empty string.
Addresses https://github.com/Irqbalance/irqbalance/issues/321