XX:-WHAT-.-FILE-.1, 5-Sep-86 15:21:21, Edit by SRA This file attempts to document the important files related to MIT host tables, where the files come from, where they go, etcetera. No warranty implied.... Which tables: The "MIT host tables" include information for the MIT chaosnet and Internet networks 18.0.0.0 and 128.52.0.0. There are also a few overflows from the Arpanet (10.0.0.0) and in past days Symbolics and CISL have also been part of the tables for various reasons. The "MIT host tables" are the authoritative source of information for the MIT Chaosnet. They also contain authoritative information for much of the relevant Internet data, but here the organization is not as clean; you should also check with the MIT Telecommunications Office, who run the MIT.EDU domain and the campus network. Contact Jeff Schiller for details. Directories: Host tables primarily live in the SYSHST directory on AI (AI:SYSHST;) and in the HOSTS directory on XX (XX:). Other machines will presumably have local copies of these. Some of the files also live in XX:; these are ones that are used directly by XX's system programs. Similarly, the binary HOSTS3 file lives in the SYSBIN directory on all ITS machines. Programs: Most relevant programs are in XX:. I used various languages as was convient (TECO, Twenex CTL, Twenex PCL, Twenex CMD, C, MIDAS). You are welcome to read this stuff but probably don't want to touch it unless you are a Real Hero. Mailing lists: The primary list for host table related issues is INFO-HOSTS@AI (aka AI.AI.MIT.EDU). There are a couple of spinoff lists. INFO-HOSTS-UPDATE@AI is a list for people who want to receive SRCCOMs of changes to HSTMIT from the nightly XX batch job. INFO-HOSTS-REQUEST is the usual list addition/removal address. NAMECALLERS@AI is a list to discuss the MIT implementation of the DARPA Domain system. You're welcome to listen in but don't speak unless you know what you are talking about. Source files: XX:HOSTS.NIC: This is a copy of the Internet host table maintained by the Network Information Center under DARPA contract. It is the closest thing available to a complete listing of the Internet, but is somewhat deficient (due to the unmanageable size of the table). The version number of this file tracks the version number of [SRI-NIC]NETINFO:HOSTS.TXT. MIT machines desiring a copy of the unadulterated NIC table should copy the one from XX rather than getting it directly from the NIC; this was an official request from the NIC, to help cut down on the load on SRI-NIC due to FTP servers. To update this file, send mail to HOSTMASTER@NIC. AI:SYSHST;HSTNET > This file contains about ten lines of machine readable data (network names and numbers for MIT nets). This part is seldom changed. Most of the file is a comment (ignored by the table compilers) listing the current subnet assignments for the MIT networks. This is the authoritative listing. No changes should be made to subnet numbers without checking this file to see if there is a conflict. There is a small self appointed militia who police this file for reasonableness; talk to the people on INFO-HOSTS@AI if you need details. AI:SYSHST;HSTLCS > This is the authoritative host table for LCS. It is dumped from the LCS LispM namespace every time a "significant" change is made to the namespace. Editing the file is a waste of time, edit the namespace instead. Format is a little odd to support some additional functionality for the LCS.MIT.EDU domain. A copy of this file lives in XX:HSTLCS.TXT. This file is also the source for the LCS.MIT.EDU domain. AI:SYSHST;HSTAI > This is the authoritative host table for the AI lab. It is dumped from the AI LispM namespace. Again, edit the namespace instead of the file. File format doesn't have the weird LCS kludges. A copy of this file lives in XX:HSTAI.TXT. This file contains the same data as the AI.MIT.EDU domain. AI:SYSHST;HSTEE > Table for the EECS "domain" (still part of MIT.EDU). Hand edited, but usually pretty up to date. Is the authoritative source for EECS Chaosnet machines. Internet addresses here have to be cleared with Telecommunications, but changes should be reflected in this file so that machines which don't grok domains yet can keep up with things. Copy in XX:HSTEE.TXT. AI:SYSHST;HSTATH > Table for the Project Athena "domain" (still part of MIT.EDU). Hand edited, may or may not be up to date at a given moment. Authoritative source for Athena Chaos machines. Copy lives in XX:HSTATH.TXT. AI:SYSHST;HSTG > Table for machines not part of LCS, AI, Athena, or EECS. Hand edited, may or may not be up to date. Authoritative for Chaosnet. Copy in XX:HSTG.TXT. AI:SYSHST;HSTXXX > This is a file of additions and kludges that need to be added to the NIC host table (some machines at Stanford and Symbolics, other things as needed). You probably shouldn't edit this. Generated files (produced by XX's nightly host table job): XX:HSTMIT.TXT Merge of HSTLCS, HSTAI, HSTATH, HSTEE, and HSTG. This has all the special kludges fixed up, names expanded, etcetera. This is the closest thing there is to a single table listing all MIT machines. Copy in AI:SYSHST;HSTMIT >. XX:HSTNIC.TXT This is derived from the NIC host table. A few known bad names and/or addresses are removed, any nicknames that conflict with MIT machines are removed. Any machines that look "insignificant" are removed (workstations at sites other than MIT, etcetera). The contents of this file are subject to change without notice as becomes necessary to feed this file into some of the later stages of the compilation process. HSTXXX is prepended to the NIC data in this file. No data for MIT machines appears in this file; it is implicitly assumed that the data in the MIT tables superceedes whatever the NIC has on file. XX:HOSTS.TXT This is HSTMIT followed by HSTNIC, but without the Chaosnet data. It is as close to a "complete" listing of the DARPA Internet as can be made with the current setup. XX uses this file for its old Internet host table software (XX also uses the new Domain system, but that's a different issue). Machines that want a copy of the NIC table with the MIT data fixed should use this file. XX:HOSTS3.TXT This is a file suitable for feeding to the HOSTS3 binary table compiler. It contains all the essential data from HSTMIT followed by HSTNIC. Some amount of pruning is done to this file to fit the address space constraints of the HOSTS3 compiler (at present, all the services entries are removed). Contents subject to change without notice. XX:HOSTS3.BIN This is the compiled binary HOSTS3 table. Many programs on ITS and Twenex use this. Copies can be found in the directory on all MIT Twenex machines and in the SYSBIN; directory on all ITS machines. Version numbers of the ITS copies track that of the XX: copy. XX:HOSTS2.TXT An old format of host table (MIDAS hackers should note the syntax!). Somewhat useless because it can only handle IP addresses with the third octet zeroed. Used to produce... XX:HOSTS2.BIN The compiled binary HOSTS2 format file. Maintained for upward compatibility. No new programs should use this. XX:HOST3C.TXT A listing of just the Chaosnet machines from HSTMIT. Name is somewhat confusing, since the file format is actually HOSTS2. I think the unix binary table compiler reads this file. Is called HOST3C because it is used to produce... XX:HOST3C.BIN A HOSTS3 binary format listing of the MIT Chaosnet. XX:MITGWS.TXT A listing of (IP) gateways and (Chaos) bridges, produced from HSTMIT. No warranty, but very handy for tracking down network problems, since it is usually more up to date than any of the gateway maps available. Copy in AI:SYSHST;MITGWS >. XX:CHAOSHOSTS.CHAOS-ALL A file in the format used by unix sendmail, usually installed as /etc/chaoshosts. This file contains a listing of all chaosnet hosts at MIT (derived from HSTMIT). XX:CHAOSHOSTS.CHAOS-ONLY Same as CHAOSHOSTS.CHAOS-ALL but only lists hosts that are on the chaosnet but aren't on the Internet. Depending on whether your vax is or isn't on the net you will want one or the other of these files so that sendmail can route mail as efficiently as possible. There are some other files associated with the LCS.MIT.EDU domain, but that's a whole separate topic and not relevant here.