i-scream documentation viewer

minutes-20001101.txt

Minutes of Meeting, 01/11/2000 @ 15:00
Location: UKC Computer Science Meeting Room

Present: ab11, ajm4, pjm2, tdb1

It was decided that we should aim to start programming in 
the Project Week, which we believe is week 8. This gives us 
a week and a half to work out what we need to do before we 
can start the implementation, and do it.

Discussion on UDP packets, and whether we really want to use
it for host to server communication. Need to find out more
about how Java implements UDP, and whether jc wants to take
this route.

ajm4 leaves after about 10 minutes, because he doesn't want 
to walk home....

It was decided that regardless of whether we used TCP or 
UDP, we would format the data sent between the host and the 
server in XML.

This is a rough list of the things we think we'll be needing 
to monitor, although some will not come until at least after 
the first main release.

Solaris
 - IP Address or Hostname
 - MAC Address
 - OS Name
 - OS Version
 
 - CPU(s) % Usage
 - Load (3 values)
 - Memory Total
 - Memory Free
 - Swap Total
 - Swap Free
 
 - Disk Space Total (per partition)
 - Disk Space Free (per partition)
 - Number of users
 
 - Open sockets (how many etc)
 - List "greedy processes" (ie. those that have consumed too 
   many resources.

Windows NT
 - IP Address or Hostname
 - MAC Address
 - OS Name
 - OS Version
 
 - CPU(s) % Usage
 - Memory Total
 - Memory Free

 - Disk Space Total (per partition)
 - Disk Space Free (per partition)

 - Shares in use
 - User connected to resources

The discussion then side-tracked on to the host to server 
communication. It was discussed that having the server 
retain all configuration ifnromation would make management 
easier. The host would know about the "well known port" that 
the Filter Manager is runing on (maybe using /etc/services). 
Also, it could be possible to use a DNS CNAME (aka wpad!) to 
locate the machine that the server is running on. This would 
mean that the host has no configuration information sotred 
locally, at least initially anyway. The configuration would 
be sent from the server to the host during the initial TCP 
communications between the host and the Filter Manager (or 
maybe initial communication with it's assigned host).

This is a very rough idea of the XML encoding we could use. 
It may not be 100% XML compliant, but that's not something 
we need to worry about right at the start. This first one 
was produced by pjm2.

<?xml version="1.0" encoding="????????" ?>
<data>
  <ip_address>###</ip_address>
  <mac_address>###</mac_address>
  <os_name>###</os_name>
  <os_ver>###</os_ver>
  <cpu>###</cpu>
  <load1>###</load1>
  <load5>###</load5>
  <load15>###</load15>
  <mem_total>###</mem_total>
  <mem_free>###</mem_free>
  <swap_total>###</swap_total>
  <swap_free>###</swap_free>
  <disk_total>###</disk_total>
  <disk_free>###</disk_free>
  <users>###</users>
  <open_sockets>###</open_sockets>
  <greedy>###</greedy>
</data>

tdb1 decided to put this forward, but doesn't have a clue 
about XML compliance... he just thinks it looks nicer. :)

<?xml version="1.0" encoding="????????" ?>
<data>
  <identity>
    <ip_address>###</ip_address>
    <mac_address>###</mac_address>
    <os_name>###</os_name>
    <os_ver>###</os_ver>
  </identity>
  <system>
    <cpu>###</cpu>
    <load>
      <1>###</1>
      <5>###</5>
      <15>###</15>
    </load>
    <mem_total>###</mem_total>
    <mem_free>###</mem_free>
    <swap_total>###</swap_total>
    <swap_free>###</swap_free>
  </system>
  <disk>
    <part>
      <name>###</name>
      <total>###</total>
      <free>###</free>
    </part>
  </disk>
  <procs>
    <users>###</users>
    <open_sockets>###</open_sockets>
    <greedy>###</greedy>
  </procs>
</data>

Things we need to do before we can commence implementation;

 - Investigate XML & Java
 - Find out about UDP, and how it works in Java
 - UML diagrams for the server and host

Meeting was concluded @ 5pm. Next meeting arranged for 
Monday, in the meeting room. Someone needs to book it soon.