Newsgroups: comp.parallel.pvm From: Josef Sikora Subject: PVM Broken Pipe Organization: Boeing Date: Mon, 4 May 1998 16:58:20 GMT Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <354DF3AC.1580DD7C@boeing.com> I am receiving the following message occasionally libpvm [t600001] mxfer() write tt sock: Broken pipe when running over a mix of SGI (SGI6) and IBM (AIX46K) workstations while running PVM 3.4b6. The connections are established using PvmRouteDirect. Probably the direct tcp/ip connection has been broken. I have the following questions. 1) Is my assumption about the broken tcp/ip connection correct and if so, what causes this to happen? 2) Does the connection reestablish itself? If it does, is the data originally sent retransmitted or is it lost? 3) Would PvmAllowDirect also have this problem? Would it also have this problem if a direct route was initially established? If the connection was reestablished, would it be direct or indirect? 4) If the error message is not related to the tcp/ip connection, what is the problem?