Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!uwm.edu!zaphod.mps.ohio-state.edu!brutus.cs.uiuc.edu!lll-winken!gauss.llnl.gov!casey
From: casey@gauss.llnl.gov (Casey Leedom)
Newsgroups: comp.bugs.4bsd
Subject: Re: inetd looping detection too short & dummy environ set wrong
Message-ID: <40987@lll-winken.LLNL.GOV>
Date: 8 Dec 89 21:49:52 GMT
References: <73423@uunet.UU.NET>
Sender: usenet@lll-winken.LLNL.GOV
Reply-To: casey@gauss.llnl.gov (Casey Leedom)
Organization: Lawrence Livermore National Laboratory
Lines: 21

| From: asp@uunet.UU.NET (Andrew Partan)
| 
| Fix:
| 	1) Increase the loop count from 40 to 400 (should hold us for a
| 	while - really need better loop detection code here).

  But this fix will leave some slower machines out in the cold.  I have to
agree with the second part of your parenthetical comment.  We really do
need a better mechanism.  Merely patching the current mechanism not only
doesn't fix the current problems, but also may aggravate them.

  A much better idea might be to look at the exit codes of the child
server processes and if more than N of the exit codes in a row were
non-zero, then fail the server.

  However this wouldn't catch a looping client which was submitting
endless legitimate requests.  Unfortunately this is an impossible
situation to detect correctly.  Legitimate (but notably low performance)
programs may perform their jobs by sponsoring hundreds of server requests.

Casey