SLAC WWW-Tech
Mtg 11/5/97SLAC WWW-Tech
Mtg 11/5/97Attendees: Tony Johnson, Les Cottrell, John Halperin, Bob Cowles, Joan Winters, Charlotte Hee, Ruth McDunn, Dennis Wisinski, Bebo White, Jennifer Masek.
Agenda:
For more information see http://www2.slac.stanford.edu/bebo/wwwtech/www6conf.htm.
It was decided that a week Friday (11/14/97) it will be moved to port 8080. Bebo will check whether there is still much interest in this site.
This issue goes back a long way. Currently the only way to look at the log files s to allow someone to logon to the server. More information can be found at http://www2.slac.stanford.edu/bebo/wwwtech/logging.htm A proposal was made by Bebo to make them available via AFS so users could roll their own analyses. Are there any privacy issues, e.g. libraries do not give out lists of what books have been checked out. There is also a possible performance issue since the files are very large. Another issue is that the files are only available after they have been closed which happens once a day. Perhaps it would be better to copy the files, this decouples the server from AFS. Then the issue came up as to why such reports are needed and what kind of information is required. One reason is for reports for the DOE to indicate how much SLAC server is used, or for groups to identify where their sites are accessed from. There are also tools now available commercially.
Bebo will look into AFS issues with systems. The Web group will put some effort into looking at analysis tools. It was agreed to buck the privacy issues to the WWW-CC.
This is a server to replace the existing primary Unix web server. There is more information available at http://www2.slac.stanford.edu/bebo/wwwtech/newuserver.htm. There are 2 servers in evaluation, Netscape Enterprise (2.0 & 3.0) & Apache 1.2.4. Bebo feels there is a need to decide which one we want to pursue. In turn this means we need to determine what are the requirements. These include: can it support SPIRES (Tony's contention is that there is a problem and he will document it), do we need SSL (already available on the www2 NT server), what about SLACONLY, do any of them support FrontPage?
People should send requirements for the new server to Bebo, so he can assimilate them and incorporate them into the evaluation.
There was a request from the Web group for 4 web server names/addresses to be run on a single NT machine and all to use port 80 and to use separate root directory addresses. The requirement is to make it easier for people to find machines, and to allow the possibility of putting the web servers on separate machines. This is a major increment in the number of holes through the screening router. We are currently using 14 of the 20 holes. About 10 of these are grandfathered. Two are supposedly due to come back soon, and another (SSRL) is due back middle of next year. The limit of 20 is twofold, one is the difficulty of administering/updating the ACLs in the screening router (the complexity increases dramatically as one adds ACLs), the second is the impact it has on router performance. A suggestion was made as to whether we could limit such servers to a set of addresses on a given subnet.
John will look at the possibility of using such a set of addresses.
Tony announced that MSIE 4.0 now runs on Solaris.
Bebo announced that sunspi.slac.stanford.edu:1234 is running the CERN server and has the same config file as thw www.slac.stanford.edu server. The idea is that in an emergency this server could provide some services if www.slac.stanford.edu takes a long term holiday.