From owner-freebsd-cluster Sat Jan 6 14:35: 3 2001 From owner-freebsd-cluster@FreeBSD.ORG Sat Jan 6 14:35:02 2001 Return-Path: Delivered-To: freebsd-cluster@freebsd.org Received: from fledge.watson.org (fledge.watson.org [204.156.12.50]) by hub.freebsd.org (Postfix) with ESMTP id 60C6F37B400; Sat, 6 Jan 2001 14:35:01 -0800 (PST) Received: from fledge.watson.org (robert@fledge.pr.watson.org [192.0.2.3]) by fledge.watson.org (8.11.1/8.11.1) with SMTP id f06MZ0720540; Sat, 6 Jan 2001 17:35:00 -0500 (EST) (envelope-from robert@fledge.watson.org) Date: Sat, 6 Jan 2001 17:35:00 -0500 (EST) From: Robert Watson X-Sender: robert@fledge.watson.org To: jasone@FreeBSD.org Cc: cluster@FreeBSD.org Subject: Re: Decomposition of "process" -- will it be possible to have unbacked procs? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: robert@fledge.watson.org Sender: owner-freebsd-cluster@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG BTW, a few cute side effects of abstracting out some of the process functionality to hide the implementation of some process functionality: - Emulated environments can expose the process structure from within an emulated operating system. For example, processes running under Windows in VMWare or plex86 could appear in the system process table -- terminate signals to the image processes could be delivered within Windows to the real process (given appropriate mapping). - The same goes for clustering environments -- when a process is on another host, or migrates, images appear on each host to reflect that change. Depending on how comprehensive the level of clustering is, operations from signaling to debugging could be forwarded. It could also be used to help support clustering in other ways -- when a process migrates, its implementor is changed so that requests originally delivered locally are now forwarded as needed. - Old-style processes and SA processes could coexist at the same time, with requests on a process being processed by different code transparently. We already do this for things like the process sysent table, some signal delivery, etc. Presumably this would be entirely compatible with the decomposition of the process such that scheduled entities are not necessarily the same as entries in the process table (not all processes that are visible are scheduled, or they may have multiple schedulable parts). Robert N M Watson FreeBSD Core Team, TrustedBSD Project robert@fledge.watson.org NAI Labs, Safeport Network Services To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-cluster" in the body of the message