<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Tahoma
}
--></style>
</head>
<body class='hmmessage'><div dir='ltr'>
Thanks i didn't know of io:format behaviour and 'user' process to redirect I/O!<br><br><div>> Date: Thu, 18 Aug 2011 11:36:37 -0600<br>> From: erlang@nimqas.com<br>> To: maggio.vincenzo@hotmail.it<br>> CC: erlang-questions@erlang.org<br>> Subject: RE: [erlang-questions] Process surviving disconnect<br>> <br>> By default a process writes to the shell which started it.<br>> <br>> Everything works as intended and expected if I change bob.erl<br>> <br>> DELETE: io:format ("Alice died of ~p.~n", [Reason] ),<br>> REPLACE: io:format (user, "Alice died of ~p.~n", [Reason] ),<br>> <br>> This way, Bob writes to his own shell and all is well.<br>> <br>> On Thu, 18 Aug 2011 15:19:05 +0200, Vincenzo Maggio wrote:<br>> > Mmm, sorry but I don't grasp it: shouldn't bob write 'Alice died of<br>> > no connection' on bob's node, which is in fact well alive?<br>> ><br>> >> Date: Wed, 17 Aug 2011 21:25:37 -0600<br>> >> From: erlang@nimqas.com<br>> >> To: maggio.vincenzo@hotmail.it<br>> >> CC: erlang-questions@erlang.org<br>> >> Subject: RE: [erlang-questions] Process surviving disconnect<br>> >><br>> >> Mystery solved: no erlang bug but a brain bug of mine.<br>> >><br>> >> When I cut the network connection, Bob tries to write "Alice died <br>> >> of<br>> >> noconnection" to Alice's console which obviously is unreachable. So <br>> >> loop<br>> >> crashEs and Bob dies.<br>> >><br>> >> On Wed, 17 Aug 2011 22:56:20 +0200, Vincenzo Maggio wrote:<br>> >> > Absolutely, maybe it's us but I notice that you, as me, has tried <br>> >> all<br>> >> > the possibilities and in fact in any case but on link <br>> >> disconnection<br>> >> > bob (and we think is right, BTW!!!) remains alive and in good <br>> >> health.<br>> >> > I hope someone shows up with a clever explanation, if not it's<br>> >> > perfectly understable to file a bug.<br>> >> ><br>> >> >> I disconnect Alice from the network -> Bob joins his ancestors.<br>> >> > Ahahah, may it rest in peace with the ones he loved!!!<br>> >> ><br>> >> ><br>> >> >> Date: Wed, 17 Aug 2011 14:49:29 -0600<br>> >> >> From: erlang@nimqas.com<br>> >> >> To: maggio.vincenzo@hotmail.it<br>> >> >> CC: erlang-questions@erlang.org<br>> >> >> Subject: RE: [erlang-questions] Process surviving disconnect<br>> >> >><br>> >> >> OK, I changed the spawn_link to spawn and subsequent monitor as <br>> >> you<br>> >> >> suggest: exactely the same behaviour.<br>> >> >><br>> >> >> Alice terminates gracefully -> Bob gets informed and is alive.<br>> >> >> Alice terminates ungracefully -> Bob gets informed and is alive.<br>> >> >> I kill Alice's node -> Bob gets informed and is alive.<br>> >> >> I disconnect Alice from the network -> Bob joins his ancestors.<br>> >> >><br>> >> >> So I will wait some days more here on erlang-questions and if<br>> >> >> nothing<br>> >> >> shows up here, I will file a bug and hopefully will be proven<br>> >> >> wrong.<br>> >> >><br>> >> >> On Wed, 17 Aug 2011 22:23:06 +0200, Vincenzo Maggio wrote:<br>> >> >> > Use monitor(process, PID_TO_BE_MONITORED) after spawn.<br>> >> >> > BTW not having two PCs I simply tested your code killing alice <br>> >> and<br>> >> >> > guess what? bob survives! :)<br>> >> >> ><br>> >> >> >> Date: Wed, 17 Aug 2011 14:13:37 -0600<br>> >> >> >> From: erlang@nimqas.com<br>> >> >> >> To: maggio.vincenzo@hotmail.it<br>> >> >> >> CC: erlang-questions@erlang.org<br>> >> >> >> Subject: RE: [erlang-questions] Process surviving disconnect<br>> >> >> >><br>> >> >> >> Thank you for your quick answer.<br>> >> >> >><br>> >> >> >> > BTW, before filing a bug, could you please substitute<br>> >> >> spawn_link<br>> >> >> >> with<br>> >> >> >> > spawn_monitor and remove the process_flag lines? It would be<br>> >> >> >> > interesting to understand if either bob dies on its own or <br>> >> it's<br>> >> >> >> > killed<br>> >> >> >> > by no more being able to communicate with alice.<br>> >> >> >><br>> >> >> >> I am not sure how to replace spawn_link with spawn_monitor, <br>> >> as<br>> >> >> >> neither<br>> >> >> >> spawn_monitor/1 nor spawn_monitor/3 take a node parameter.<br>> >> >> >> How do I do that or how else can I get some more detailed<br>> >> >> >> information<br>> >> >> >> about Bob's sudden passing the Styx.<br>> >> >> >><br>> >> >> >><br>> >> >> >><br>> >> >> >> On Wed, 17 Aug 2011 21:48:30 +0200, Vincenzo Maggio wrote:<br>> >> >> >> > Hello,<br>> >> >> >> > I think something is wrong here.<br>> >> >> >> >> Bob die of noconnection.<br>> >> >> >> ><br>> >> >> >> > This is printed by<br>> >> >> >> >> {'EXIT', Bob, Reason} -><br>> >> >> >> >> io:format ("Bob died of ~p.~n", [Reason] ),<br>> >> >> >> ><br>> >> >> >> > So alice is in fact receiving bob last death cry :D and<br>> >> >> >> process_flag<br>> >> >> >> > translate it in a message instead of transmitting exit <br>> >> signal<br>> >> >> to<br>> >> >> >> > alice; I think this is ok from the point of view of Alice, <br>> >> so<br>> >> >> the<br>> >> >> >> > real<br>> >> >> >> > problem is that bob is dying (I know it's mundane, but I<br>> >> >> learned<br>> >> >> >> not<br>> >> >> >> > to make assumption).<br>> >> >> >> > Mmm, well I don't know if having no more a connection <br>> >> between<br>> >> >> the<br>> >> >> >> > process makes Erlang VM do some assumption of a virtual <br>> >> master<br>> >> >> >> node.<br>> >> >> >> > Well, if you want my opinion, I think that you should file a<br>> >> >> bug<br>> >> >> >> on<br>> >> >> >> > the Erlang bugs mailing list if no one comes up with a <br>> >> proper<br>> >> >> >> > explanation.<br>> >> >> >> > Even if what we're thinking is wrong and this is not a bug, <br>> >> in<br>> >> >> the<br>> >> >> >> > past I had a problem on node lookup and they resolved it.<br>> >> >> >> > These are my two cents, but if you can please let me know if<br>> >> >> there<br>> >> >> >> > are further updates 'cause it's a really interesting <br>> >> problem.<br>> >> >> >> ><br>> >> >> >> > BTW, before filing a bug, could you please substitute<br>> >> >> spawn_link<br>> >> >> >> with<br>> >> >> >> > spawn_monitor and remove the process_flag lines? It would be<br>> >> >> >> > interesting to understand if either bob dies on its own or <br>> >> it's<br>> >> >> >> > killed<br>> >> >> >> > by no more being able to communicate with alice.<br>> >> >> >> ><br>> >> >> >> > Vincenzo<br>> >> >> >> ><br>> >> >> >> >> Date: Wed, 17 Aug 2011 10:57:35 -0600<br>> >> >> >> >> From: erlang@nimqas.com<br>> >> >> >> >> To: maggio.vincenzo@hotmail.it<br>> >> >> >> >> CC: erlang-questions@erlang.org<br>> >> >> >> >> Subject: RE: [erlang-questions] Process surviving <br>> >> disconnect<br>> >> >> >> >><br>> >> >> >> >> Thank you very much Vincenzo. You affirmed my assertion <br>> >> that<br>> >> >> Bob<br>> >> >> >> >> should<br>> >> >> >> >> survive the disconnect.<br>> >> >> >> >> Nevertheless he dies.<br>> >> >> >> >> I will point out exactly what I do and maybe someone can <br>> >> spot<br>> >> >> >> the<br>> >> >> >> >> error<br>> >> >> >> >> in my code, my setup or my thinking and tell me what I am<br>> >> >> doing<br>> >> >> >> >> wrong.<br>> >> >> >> >><br>> >> >> >> >> I start a node on gca.local:<br>> >> >> >> >> unroot@gca:~$ erl -name 'bob@gca.local' -setcookie 123<br>> >> >> >> >><br>> >> >> >> >> I start a node on usa.local:<br>> >> >> >> >> unroot@usa:~$ erl -name 'alice@usa.local' -setcookie 123<br>> >> >> >> >><br>> >> >> >> >> I start sasl on bob@gca.local:<br>> >> >> >> >> (bob@gca.local)1> application:start (sasl).<br>> >> >> >> >><br>> >> >> >> >> I run alice:start/0 on alice@usa.local:<br>> >> >> >> >> (alice@usa.local)1> alice:start ().<br>> >> >> >> >> true<br>> >> >> >> >><br>> >> >> >> >> I look for bob on bob@gca.local and save its pid:<br>> >> >> >> >> (bob@gca.local)2> whereis (bob).<br>> >> >> >> >> <0.65.0><br>> >> >> >> >> (bob@gca.local)3> Pid = whereis (bob).<br>> >> >> >> >> <0.65.0><br>> >> >> >> >><br>> >> >> >> >> I cut the network cable and wait a minute for the timeout.<br>> >> >> >> >><br>> >> >> >> >> On alice@usa.local I get the following output:<br>> >> >> >> >> =ERROR REPORT==== 17-Aug-2011::10:53:21 ===<br>> >> >> >> >> ** Node 'bob@gca.local' not responding **<br>> >> >> >> >> ** Removing (timedout) connection **<br>> >> >> >> >> Bob die of noconnection.<br>> >> >> >> >><br>> >> >> >> >> Nice, Alice trapped Bob's death and reported it. I check <br>> >> for<br>> >> >> >> Alice:<br>> >> >> >> >> (alice@usa.local)2> whereis (alice).<br>> >> >> >> >> <0.42.0><br>> >> >> >> >><br>> >> >> >> >> Alice is up and running.<br>> >> >> >> >><br>> >> >> >> >> On bob@usa.local I get the following output:<br>> >> >> >> >> =ERROR REPORT==== 17-Aug-2011::10:53:10 ===<br>> >> >> >> >> ** Node 'alice@usa.local' not responding **<br>> >> >> >> >> ** Removing (timedout) connection **<br>> >> >> >> >><br>> >> >> >> >> But Bob is dead:<br>> >> >> >> >> (bob@gca.local)4> whereis (bob).<br>> >> >> >> >> undefined<br>> >> >> >> >> (bob@gca.local)5> is_process_alive (Pid).<br>> >> >> >> >> false<br>> >> >> >> >><br>> >> >> >> >> I really do not understand what is happening.<br>> >> >> >> >><br>> >> >> >> >><br>> >> >> >> >> Thank you in advance<br>> >> >> >> >><br>> >> >> >> >> Anchise<br>> >> >> >> >><br>> >> >> >> >> Here goes the code I used:<br>> >> >> >> >><br>> >> >> >> >> -module (alice).<br>> >> >> >> >> -compile (export_all).<br>> >> >> >> >><br>> >> >> >> >> start () -> register (alice, spawn (fun init/0) ).<br>> >> >> >> >><br>> >> >> >> >> stop () -> whereis (alice) ! stop.<br>> >> >> >> >><br>> >> >> >> >> init () -><br>> >> >> >> >> process_flag (trap_exit, true),<br>> >> >> >> >> Bob = spawn_link ('bob@gca.local', bob, start, [self () ] <br>> >> ),<br>> >> >> >> >> loop (Bob).<br>> >> >> >> >><br>> >> >> >> >> loop (Bob) -><br>> >> >> >> >> receive<br>> >> >> >> >> stop -> ok;<br>> >> >> >> >> {'EXIT', Bob, Reason} -><br>> >> >> >> >> io:format ("Bob died of ~p.~n", [Reason] ),<br>> >> >> >> >> loop (Bob);<br>> >> >> >> >> Msg -><br>> >> >> >> >> io:format ("Alice received ~p.~n", [Msg] ),<br>> >> >> >> >> loop (Bob)<br>> >> >> >> >> end.<br>> >> >> >> >><br>> >> >> >> >><br>> >> >> >> >> -module (bob).<br>> >> >> >> >> -compile (export_all).<br>> >> >> >> >><br>> >> >> >> >> start (Alice) -><br>> >> >> >> >> process_flag (trap_exit, true),<br>> >> >> >> >> register (bob, self () ),<br>> >> >> >> >> loop (Alice).<br>> >> >> >> >><br>> >> >> >> >> loop (Alice) -><br>> >> >> >> >> receive<br>> >> >> >> >> stop -> ok;<br>> >> >> >> >> {'EXIT', Alice, Reason} -><br>> >> >> >> >> io:format ("Alice died of ~p.~n", [Reason] ),<br>> >> >> >> >> loop (Alice);<br>> >> >> >> >> Msg -><br>> >> >> >> >> io:format ("Bob received ~p.~n", [Msg] ),<br>> >> >> >> >> loop (Alice)<br>> >> >> >> >> end.<br>> >> >> >> >><br>> >> >> >> >><br>> >> >> >> >> On Wed, 17 Aug 2011 12:47:02 +0200, Vincenzo Maggio wrote:<br>> >> >> >> >> > Hello,<br>> >> >> >> >> > without further info a debug is rather difficult.<br>> >> >> >> >> > But let's try to at least start analysis of the problem:<br>> >> >> >> >> ><br>> >> >> >> >> >> - Has this something to do that I initially spawn Bob<br>> >> >> from<br>> >> >> >> the<br>> >> >> >> >> >> Alice<br>> >> >> >> >> >> node?<br>> >> >> >> >> ><br>> >> >> >> >> > Absolutely not: this would hit the very foundation of<br>> >> >> Erlang,<br>> >> >> >> >> process<br>> >> >> >> >> > referential transparency. When a process is started is a<br>> >> >> brand<br>> >> >> >> >> new,<br>> >> >> >> >> > clean entity (indeed, default process heap space is <br>> >> always<br>> >> >> the<br>> >> >> >> >> same<br>> >> >> >> >> > size!).<br>> >> >> >> >> ><br>> >> >> >> >> >> - How can I make Bob to survive a connection loss?<br>> >> >> >> >> ><br>> >> >> >> >> > Look above: it SHOULD survive.<br>> >> >> >> >> ><br>> >> >> >> >> > Can you please start SASL (application:start(sasl) from <br>> >> the<br>> >> >> >> shell)<br>> >> >> >> >> > and see if shell log puts some further information?<br>> >> >> >> >> ><br>> >> >> >> >> > Vincenzo<br>> >> >> >> >><br>> >> >> >><br>> >> >><br>> >><br>> <br></div> </div></body>
</html>