RE: issue: parallel rsh'd makes on >=2GHz P4's stalling

[Hayden Ridenour <hridenou at interwoven dot com>]

This one has me stumped.  We've been using Cygwin for building some sizable
source sets for over two years, starting with Cygwin-B20.1.  We're currently
running 1.3.6.  We've also tried 1.3.9.  Cygwin has been successfully
handling parallel rsh jobs (first with the help of RSHD from Denicomp, then
using inetutils & rshd when that was ported) for that whole time period on
several different PC models.

Now, these jobs are stalling on new Dell Precision Workstation 340's and
530's with Intel's new 512MB cached P4's (2-2.2GHz).  Our older systems,
including a pair of 530's with 1.7GHz, 256MG cached P4's continue to work
fine.  The stalling only occurs when we get to a section of the build
process that goes parallel.  The single-tasking sections continue to run
reliably.  The stalling includes either two or three parallel processes,
each a recursive make, that operate for a while and then go into loops.
Each process consumes as much CPU as it can get.  None of the processes
families coming off of in.rshd are showing I/O blocked status.

If we kill a process, the rest of that recursive make stream continues on as
does the other rsh'd process sets.

We're working on creating a tighter test case to investigate the behavior
but I wanted to see if anyone here has run into this issue.  A search of the
archives didn't show any similarly reported cases but I'm hoping someone
else might have seen this or have some ideas as to what could be causing
this.

    Hayden

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/