question about policy daemon time limit
in [Postfix]
|
From: Egoitz Aurrekoetxea Aurre on 13 Sep 2009 08:58 >> >> When you run the policy service under the Postfix spawn daemon, >> terminate when a read operation returns an end-of-file indication. >> >> Wietse > > Hi Wietse, > > Have just read this... yep this is just what I do, I read from input > fd with <> wich in perl does the trick. I didn't remember I exited > from reading loops with this, because I ended writting this code > perhaps one year ago... and from there has been in testing proccess > although I haven't looked almost at the code because no errors have > appeared. So... it's done then, all perfect :). > > > Bye thanks!!! Although I simulated yesterday the pocily_time_limit exceed.... I set 40 seconds for policy time limit and too for smtpd_policy_service_max_ttl.... I telneted the smtpd later after data command I waited 2 minutes (for exceeding nice the 40 seconds) for sending end of data... the mail policy was queried anyway for the result and yes when I type quit for closing connection policy too received end-of-file and exited... just when I closed.. so seems in postfix 2.6 if policy time is less than you needed for delivering mail seems that Postfix maintains it running till the actual mail is delivered. I have simulated too without entering data command just with recipients to see postfix behavior and it's the same... I have checked it in 2.6 don't know what could happen in previous versions... Thanks a lot!!!!!
From: Wietse Venema on 13 Sep 2009 09:22 Egoitz Aurrekoetxea Aurre: > >> > >> When you run the policy service under the Postfix spawn daemon, > >> terminate when a read operation returns an end-of-file indication. > >> > >> Wietse > > > > Hi Wietse, > > > > Have just read this... yep this is just what I do, I read from input > > fd with <> wich in perl does the trick. I didn't remember I exited > > from reading loops with this, because I ended writting this code > > perhaps one year ago... and from there has been in testing proccess > > although I haven't looked almost at the code because no errors have > > appeared. So... it's done then, all perfect :). > > > > > > Bye thanks!!! > > Although I simulated yesterday the pocily_time_limit exceed.... > > I set 40 seconds for policy time limit and too for > smtpd_policy_service_max_ttl.... I telneted the smtpd later after data > command I waited 2 minutes (for exceeding nice the 40 seconds) for Perhaps surprisingly, Postfix starts a new policy daemon when a policy daemon process terminates too soon. This means that Postfix is robust for such mistakes. It is more efficient, however, when the policy daemon waits until it is time to exit. In particular, exiting after each query would be horrible inefficient. Wietse
From: Egoitz Aurrekoetxea Aurre on 13 Sep 2009 09:33
>> >> >> I set 40 seconds for policy time limit and too for >> smtpd_policy_service_max_ttl.... I telneted the smtpd later after >> data >> command I waited 2 minutes (for exceeding nice the 40 seconds) for > > Perhaps surprisingly, Postfix starts a new policy daemon when a > policy daemon process terminates too soon. > > This means that Postfix is robust for such mistakes. Yep should be said is very very nice... I come from Qmail wich IMHO it's too a nice MTA but ... since I used Postfix I haven't set up any other mta... just Postfix lol (in fact I have written PQR)!! although I should say I like qmail too. > > It is more efficient, however, when the policy daemon waits until > it is time to exit. It does :) > In particular, exiting after each query would > be horrible inefficient. Yep the policy just dies on EOF, or well if you by hand enter a blank line without entering before in that telnet session request=smtpd_access_policy... so :) it's well. > > > Wietse thankzzz!!! |