Version 1.5 release. Most significant changes listed below:

• Fixed bus:
  
  
  • Small bug with "flowd -k confcheck" fixed. flowd was unable to check configuration if no running copy of flowd exists. Thank to Inozemtzev Paul for report.
    
    
  • Router uptime bug fixed. If routers uptime reach 2^32 msec (approx. 50 days) flowd stops netflow processing for those router.
    
    
  • Linux Redhat9 and FreeBSD 5.* installation troubles fixed.
    
    
  • Flowd startup coredump bug fixed. Thank to Pavel E. Sabiriyanov for report.
    
    
  • Flow cache size change and "flowd -k reconfigure" core dump bug fixed. Thank to Leonid Vasilev.
    
    
• ISP billing:
  
  
  • Detail report generation option is available for customers.
    
    
  • Split traffic direction option is avalable for traffic zones. It allows you splited in to two subreports customers detail reports for specified traffic zone, for ech billing direction, incoming and outgoing. Billing traffic direction (don't confuse with real interface traffic direction) calculates by the loader program on the assumption of interface traffic direction (In, Out) and interface type (Client, External, Internal).
    
    
  • Online help for form fields.
    
    
• Autostart loader feature. If 'loader_autostart' option set to on in flowd.conf (it set to on by default) then flowd automatically start loader program after dump flowc cache. Periodical runs loader via cron now is unnecessarily. Depends on flowd.conf 'parallel_loading' option settings (off or on) single or multiply loaders (one loader for each router) will be started.
  
  
• Old ASCII address storage in MySQL is not supported. Old style flow tables rotate (flows -> flows_old) is not supported. Traffic counters, ports, AS and protocols accounting enabled always.
  
  
• If you transit router, you can define in sngle line that all interface are external using '*' as interface name (see the src/flowd.conf example).
  
  
• Improved performance, all linked lists flowd structures replaced by arrays. It a first step toward high performance netflow accounting processing.
  
  
• Per router flow cache autodump. In previous versions if one routers flow cache utilization reach 98%, all routers flow caches will be automatically flushed to dump file.
  
  
• New port/protocol counter implemented. You have ability to track in real time traffic for different TCP/UDP ports and protocols. Graphical counter presentation also available. See the src/flowd.conf example for use.
  
  
• Custom report form was significantly enhanced.
  
  
• Embeded graphical subsystem. You can use graphics for tracking traffic counters, protocol counters or flowc monitoring via flowd created graph binary files. Graphics drawing performs by PHP scripts, without using RRD Tool. RRD Tool by default is impossible to correct display flows older than 5 min. RRD Tool is not needed now, but you need to compile PHP with graphics support (libgd and libpng).
  
  
• "ignore_null_iface" option added to flowd.conf. If it turn to "on" all packets destined to Null interface (for example disabled by access-lists) will be ignored by flowd. By default it turn "on".
  
  
• Dump files with size more than 2Gb correct displays by loader_stat.php. Thank Denis Burduk for idea. If you compile PHP with safe_mode option, you need to copy /bin/ls in to PHP safe mode exec directory.
  
  
• --enable-aggregation configure option is deprecated. Instead, 'ports_aggregation' option in flowd.conf is available on per router basis. see src/flowd.conf example.
  
  
• Logging to file without using syslog. Default log file is /var/log/flowd.log.
  
  
• loader improved performance. Most of this options increase risk of data lost or dublication in case of MySQL server instability and must be used carefuly. For detail description read 'Optimization' section in Flowc manual.
  
  
  • 'update_offset N' option allows update routers table after N inserted flow records. In old releases routers table updated each time after inserting flow record. Thank Egor Egorov and Denis Burduk for idea.
    
    
  • 'delayed_insert' option allows loader use INSERT DELAYED statements instead standard INSERT.
    
    
  • 'bulk_insert' option allows loader use bulk inserts in to flows* tables via temporary tables. With this option rebuilding indexes in flows table occures once. This techniques improve loader performance to 30 times on large flows tables (100 millions rows and more) and improve disk subsystem utilization. Thank to Egor Egorov and Denis Burduk for idea and first implementation.
    
    
  • loader use read buffer. In old releases loader reads one flow record from data file before inserting in MySQL database. This feature improve disk subsystem utilization.
    
    
  • 'parallel_loading' option allows multiply loaders (one loader for each router) running.