Autor: Graeme Fowler Datum: To: Exim-users Betreff: Re: [exim] exim4 only queues mails sent by systemd service
On 16 Oct 2018, at 18:03, Ian Zimmerman via Exim-users <exim-users@???> wrote: > I think this is a misunderstanding both on the part of Lennart and on
> the part of Graeme.
I’m not so sure that it is; I was addressing the very specific example given.
> The example systemd unit shown by the OP was just _a test_. The real
> use case with a problem (or at least a potential problem) is not a human
> user sending mail with mutt or /bin/mail or whatever, but a service
> sending mail as a means of communicating status. Sometimes such
> services have no run time configuration item to change how mail is
> sent. I think this is the case with Vixie cron as used in Debian, even
> today.
Services spawned from systemd that are long-running don’t suffer this issue. It’s the one-shot units that do, and only those, so far as I can see due to the way systemd reaps them. It simply doesn’t account for a process which may have spawned other background processes when it runs.
> One could argue that the bug is in the service, and I'd tend to agree.
The bug is in the interaction between the choice of init (systemd), the options in the unit file, the configuration of mail (or rmail, or mailx, whichever is being used), the Exim command line, and the Exim configuration. There’s not really a single entity which causes the issue.
There are ways to work around the issue, as I described previously. Making Exim not do instantaneous delivery is but one of them.