Re: [Fetchmail-users] delay interval

[Carlos E. R. <rob...@te...>]

On 2017-01-08 13:41, Matthias Andree wrote:
> Am 06.01.2017 um 18:00 schrieb piequiex:
>>> El 2017-01-04 a las 17:24 -0000, piequiex escribi??:


>>> Probably your script is exiting too early, before it finishes processing.
>> Yes. External application, executed from a script, sometimes terminates with
>> the strange error - "Cannot allocate memory". This does not happen in example
>> above.
> Getting error handling right in procmail requires jumping through hoops
> and adding an error trapping recipe after each and every action that you
> code, that's why I'm advocating replacing procmail.

Yes, procmail is definitely not easy to trace, but I use log entries to
help. After years of familiarity and that my setup "simply" works, one
finds no need to invest many days in changing it ;-)

Eventually I may have to change, but meanwhile...

> While you're writing in a later message that you've found the cause for
> the allocation errors, I still believe that the error non-handling in
> your procmailrc setup persists. Reason why I think that: Fetchmail
> 6.3.26 in default configuration will leave mail on the server and retry
> it later (--softbounce is default), but for that to happen it needs to
> know that the --mda failed, either by it crashing, or from its nonzero
> exit status. Since you wrote about mail loss, I believe that procmail
> hasn't propagated that script failing status back to fetchmail.

I have postfix in the mix, so mail is queued inside postfix. It helps.

One trick is that I make procmail to copy every mail to a backup folder.
Thus If something bad happens later on, the posts are not lost. So far I
had no need.



-- 
Cheers / Saludos,

		Carlos E. R.
		(from 42.2 x86_64 "Malachite" at Telcontar)