
jesus at omniti
May 31, 2002, 1:55 PM
Post #4 of 7
(1352 views)
Permalink
|
|
restarting automaticaly the Wackamole service
[In reply to]
|
|
Yair Amir wrote: >Hi Theo, > >I really am not sure what the problem is. > >I run Spread and immediately I run a client program (sptuser) that >connects and joins a group. > You don't want that trigger that behaviour with wackamole. If Spread starts and hasn't performed a daemon membership yet (comes up alone at first), then wackamole will start, see no peers and then assume all the IP address which will confuse everyone tremendously. You really only want wackamole to work on stable, settled Spread rings. Hence the 10 seconds -- which works "most" of the time for me. On some machines I set it up to 1 minute. >Spread takes a few seconds to complete the membership, but once it >does, the client gets a response for the join. Is your case different? > On some of my machine it can take 15 - 30 seconds to complete the daemon memebership join. >In general, forking, setsid, etc. are things I would usually avoid if not >necessary in order to keep the code simple and portable to any kind of >environment (not only unix). > > All of the code now "knows" if it is unix by arch.h. And the autoconf patches I supplied make it very easy to use setsid and fork only if they are available. Spread is supposed to be a daemon... Like it or not, daemons on unix should setsid and fork... it's just they way things are -- as dictated by Stevens ;-) Otherwise the process is susceptable to receiving unwanted signals from its parent process and has a controlling terminal (which it should not). -- Theo Schlossnagle Principal Consultant OmniTI Computer Consulting, Inc. -- http://www.omniti.com/ Phone: +1 301 776 6376 Fax: +1 410 880 4879 1024D/82844984/95FD 30F1 489E 4613 F22E 491A 7E88 364C 8284 4984 2047R/33131B65/71 F7 95 64 49 76 5D BA 3D 90 B9 9F BE 27 24 E7
|