1. 02 Jul, 2008 1 commit
    • 's avatar
      Endpoints are now named "subjects", each subject consists of one host for... · 81332213
      Endpoints are now named "subjects", each subject consists of one host for incoming and one host for outgoing traffic. Host for incoming traffic can be the same as the one for outgoing traffic anyhow...
      Lots of "ifdef's", why only I haven't already branched this nice app :)
      Important memory-corruption fix.
      Added graph for matched flows.
      Improvements in GUI, now each subject has its own color everywhere/anytime.
      Tables stylesheet for gui (dirty hack).
      Modularization of graph.php started.
      git-svn-id: file:///home/svn/mapi/trunk@1413 8d5bb341-7cf1-0310-8cf6-ba355fef3186
  2. 21 Feb, 2008 1 commit
    • 's avatar
      packetloss: · 84f4fca2
       for new gui added cofiguration file and config reader (beta but working)
       everything should be backward-compatible with original commandline arguments
       if NEWGUIONLY is not defined
       configuration in file: hosts can be disabled, max. number of flows
       can be limited, timestamp difference among host can be tuned 
       reconnection can be turned on on-demand (SIGUSR1, NEWGUIONLY) becouse
       if trying every loop it seems to be more time consuming than acceptable
       reconnection was not tested with --enable-reconnect mapi mechanism as
       it was originally based just on retrieving NULL from mapi_read_result,
       however it should work as I modified it with respect to previous commit
       from AK
       stats can be cleared (SIGUSR2)
       unified reporting of application state
       commandline options (NEWGUIONLY)
       enabled use of rrd server
       improved error reporting
      git-svn-id: file:///home/svn/mapi/trunk@1384 8d5bb341-7cf1-0310-8cf6-ba355fef3186
  3. 13 Nov, 2007 1 commit
  4. 24 Oct, 2007 1 commit
    • 's avatar
      Added support for multiple host, which includes multiple interface support too. · 4ff73f0d
      Tested with new GUI (PHP); CGI GUI shows just first two hosts.
      Flow is assigned to opposite sensor when it appears there too (same flow).
      Unassigned flows are dropped, so filter has not to be exact,
      something void like "greater 0" should work.
      The differences against prev. impl. of multiple interface support
      (subject to keep or change?):
      Usage: hostA:if1:if2, hostB... "filter"
      Usage: "hostA:if1, hostA:if2" "hostB..." "filter"
      -more typing
      +DiMAPI syntax
      ?no need to parse and store device names
      mapi_stats("hostA:if1", &(stats1->statstable[i%countA]))
      mapi_stats("hostA:if2", &(stats1->statstable[++i%countA]))
      mapi_stats("host1:if1, host1:if2", &host[0].mapi_stats_dev).
      -order of devices not guaranteed so reordering* is done via pointers
      +less mapi function calls
      ?no need to store device names
      ?unique interface names required in case of reordering*
      *)for now, order never changes
        for(i = 0; i < stats1.count;i++) { // host1
          stats = stats1.statstable[i];
        for(i = 0; i < stats2.count;i++) { // host2
        for(h = 0; h < hosts; h++) {
          for(d = 0; d < host[h].devices; d++) {
            stats = host[h].mapi_stats_dev[d];
      parse_hosts() -- not a function (more simple as device names are not parsed and
      stored, just counted)
      get_mapi_stats() -- not a function (more simple because of prev. mentioned diffs)
      GUI (CSV, PHP, JS) -- several updates
      git-svn-id: file:///home/svn/mapi/trunk@1340 8d5bb341-7cf1-0310-8cf6-ba355fef3186
  5. 14 Oct, 2007 1 commit