Ping Plotter Revision History
Latest version, PingPlotter 2.60 released February 14th, 2005February 14th, 2005. PingPlotter 2.60
This release has a number of new features, and a few non-feature changes (ie: bug fixes).
New Features
- PingPlotter now supports TCP traceroute, in addition to UDP and ICMP. This allows additional options in tracing to a site that's using a firewall blocking UDP and ICMP. For this release, TCP is supported under Windows 2000, XP and 2003 only. For users of Windows XP SP2 (which blocks the sending of TCP raw sockets), we support the use of WinPcap to create the TCP packets.
- New "Getting started" screen with links to help content.
- Significant improvements in ability to run as a non-administrative user. Admin user can set default setups and non-admin users "inherit" those settings, but then have their own setups (stored in their local profiles). You still need to be an administrator to install PingPlotter, though.
- All new options dialog that includes new features for version 2.60, in addition to having a more intuitive layout and "look and feel".
- When an alert is triggered on a specific hop, that hop shows an indicator, with a float-over hint of which alert is firing.
- Selecting a hop in the upper graph highlights it in the time graph (if it's shown there).
- Alert email body text can be customized (a huge benefit for alerting to a mobile phone / SMS / etc, where the current alert message is too long).
- You can now enter a time as a "Samples to Include". PingPlotter will try and calculate how many samples will equal that time period. Example: enter "30" minutes. If you have a 2.5 second trace interval, PingPlotter will convert this to 720.
- PingPlotter now tells you about new versions with a dialog with options to supress, rather than just showing that an update is available on the status bar.
Other changes
- Changed the way packet queuing works on all packet types to reduce the number of "extra" packets that get sent out after the final destination is reached.
- Interhop delay setting is now honored more rigorously (especially on UDP and TCP types).
- When PingPlotter was running, rebooting was sometimes blocked until PingPlotter was closed.
- Float-over hints sometimes got cut off - they now make themselves fit onto the screen
- Right-clicking a time-graph in the label regions didn't pop up the menu. Fixed.
- Testing an alert email now pulls in the current final destination to flush out email body
- Changing "Max samples to hold in memory" from a non-zero to all (0) would delete all samples in memory.
- On multi-monitor systems, if PingPlotter was closed while visible on a secondary monitor, and then restarted with that monitor disabled, it would be shown off-screen. This has been corrected.
- Applying changes in the options dialog now saves the configuration immediately, rather than waiting till PingPlotter is closed next.
- Misc tweaks / tidbits.
June 16, 2004. PingPlotter 2.51.1
This is primarily a bug fix.
Bug Fixes
- When "# of times to trace" was set to 1, hitting the "Start" button wouldn't update the button with the proper status.
- Users without administrative rights could have PingPlotter crash when launching.
- Users without administrative rights would have problems with license key being recognized in some cases.
- List index out of bounds error on some route changes - fixed.
- Access violation when right-clicking the main window when no trace was running - fixed.
Other Changes
- Route change mask can now be auto-applied by multiselecting several routes and right-clicking the appropriate hop. Only the selected routes will be used in the calculation.
May 25, 2004. PingPlotter 2.51
New Features
- Allow configuration of SMTP server port, in addition to validated username / password logins to the SMTP server for sending alert emails.
- Targets that are associated with an alert are now listed on the alert configuration screen.
- Multi-selecting routes in the route-change window now highlights the hops that have changed in the upper graph area. This makes it a lot easier to find which hops are oscillating so a route change mask can be added.
- All errors now use an error dialog with a "Details" option, which shows more details (primarily, the stack trace) about an error that can be emailed to support. Additionally, the support email address and support page URL are listed.
Bug Fixes
- Test sending an alert email would leave a thread attempting to redeliver that email if the initial test failed.
- List Index Out of Bounds error could happen during "aggressive" route changes.
- Access Violations could sometimes occur if right-clicking a hop just when a route change made that hop obsolete.
- Route discovery logic was too aggressive when high packet loss was on the final destination. Route would lengthen on some occasions and show lower packet loss than it should.
- Route wouldn't "shorten" once it was lengthened due to a poorly responding final destination. This could leave one or more blank hops before the final destination, and the final destination would never recover these hops.
- Errors were not being logged to the PingPlotter.rip file correctly, making bugs much harder to track down.
- Automatically generated filenames (alerts, auto-save names) didn't allow non US characters.
- $host and $dest in alerts would always used the same name, even if the alert wasn't targeted at the final destination.
- Some .pp2 save files would not be reloadable if the save file had more hops than the current configuration allowed.
- Resuming PingPlotter when data was sent from MultiPing didn't always correctly trace the full route - very specific instances only (it's unlikely that anyone was significantly impacted by this).
- Log to file alert always wrote the current time rather than the actual sample time.
Other Changes
- Changed the way alerts are associated with targets when saved and reloaded. This means you'll need to re associate your alerts to the targets they should be watching. The alerts configuration is all saved, you'll just need to re-tell PingPlotter which targets you want to watch.
- Some shared engine code changes made in MultiPing was merged back in to PingPlotter.
- Minor "tweaks" to better accommodate users running large fonts (primarily in the options dialog).
- A couple of minor broken links in the help file were corrected.
March 13th, 2004 - Version 2.50
Version 2.50 has several significant new features and changes since 2.41, including:
New Features
- Now supports UDP (ie: Unix-Style) packets. This should allow PingPlotter to be useful in some network conditions where ICMP is being blocked. Under NT/2000/XP, you must be logged in as a administrator to send UDP packets.
- DSCP / TOS byte of packets can now be changed for better testing of VOIP (and other QoS enabled) networks.
- Integration with MultiPing - MultiPing can now send data to PingPlotter (for saving, image creation, etc). MultiPing can also launch PingPlotter and start tracing a target from the right-click menu.
- Improved route change handling - especially when hop N-1 (where N = the final destination) times out 100% of the time.
- Adding a route mask from the right-click window now merges the data for those hosts immediately so there is no need to reset that trace session.
- New INI file setting to bind to a specific adapter (works with Raw Socket and UDP only). Let us know if you want to use this feature and we will send instructions.
- New logo and application icons.
- Internal engine errors now show error message in a panel above graph to help troubleshooting problems. This data is transient, and is lost when the next error occurs or when the application is closed.
Bugs fixed since release 2.41
- Fixed a floating point divided by zero error when sizing "just" right (and small).
- Route change exclusions weren't being loaded correctly in the 2.40/2.41 releases.
- Large fonts work much better - previous 2.4x releases had unusable Alert dialog with large fonts, and would cut off text on the graphs in the main window.
- .pp2 (save) files would sometimes not reload correctly if route was only one hop long in save file.
- Error would sometimes occur when "Samples to Include" was increased and focus was set on period slightly prior to current.
- Additional "WhoIs" lookups listed on menu, but didn't do actual lookup. (See the tutorial for instructions on this)
Other changes
- The packet sizes specified in 2.41 and earlier was used for the *cargo* size, which meant that the actual packet size was an additional 28 bytes. In this release, the specified packet size is the actual TCP/IP packet size (cargo + header).
- The 2.41 and earlier default packet cargo (Packet Number + $AA) was recognized by Snort as an attack by a ICMP PING Cyberkit virus/worm. The default packet has been changed to Packet Number + "PingPlotter" + Version Number. In addition to not firing the Snort rule, this is more polite as network admins can tell what program is sending the packets, and act accordingly.
- Allow file names in auto-save image and auto-save data to include seconds ($second).
- Changing "Maximum samples in memory" immediately trims data if sample count exceeds setting (previously, you had to trace at least once to trim the data)
June 30th, 2003 - Version 2.41 released
Version 2.41 is a relatively minor update and has the following new features and bug fixes (since 2.40):
New Features
- Because of the recent expiration of the Unisys license that affected GIF files, we now support creation of GIF images, in addition to PNG and BMP.
- PNG Images now default to 256 color, resulting in about a 25% size savings (previously, the only format was 24 bit color on PNGs). An INI switch allows setting back to 24 bit color instead of 8 bit. E-mail to support@pingplotter.com for details on how to use this.
- The auto-save configuration dialog now has buttons for "Save Now", to allow for testing of switches, as well as being able to do an immediate save whenever needed. Note that doing a "Save Now" in no way affects the point at which the next auto-save happens.
Bugs Fixed
- WHOIS Window has a popup error on Windows 95/98/ME. This error was non-critical (clicking it continues fine). FIXED.
- No way to turn off the time graph for a destination that wasn't reachable. With 2.41, right-clicking on the time graph allows you to hide it.
- Right-clicking on "Destination unreachable" line in upper graph now acts like it was the final destination - allowing WhoIs lookups, alert setup, and showing / hiding of timeline graphs.
- Log to file alert writing all data instead of just data when alert conditions are matched. FIXED.
- Ignore first hops settings worked, but wasn't saved between sessions. FIXED.
- Some dual processor machines would go to our license server each time Ping Plotter is run. We don't want to know how often you run Ping Plotter, and this is a bug. FIXED.
- Auto-save frequency wouldn't allow use of "Days" and converted that to seconds instead. Note that the maximum interval is roughly 35 days.
- Certain command line switch combinations indicated that the Pro Version was necessary. FIXED.
March 9th, 2003 - Version 2.4 released
Version 2.40 has the following new features and bug fixes (since 2.30.1):
- Complete re-write of internal model. Allows some capabilities in this version and lots of cool stuff in the future.
- Annotations can be tied to a point in the time graph to show notes. This gets saved in all images.
- Save files format was changed and now takes up only 5-15% of what it used to (much of this is due to new internal compression).
- Entering a full URL in the trace address box now interprets correctly and uses only the base address automatically.
- Can now "Copy as summary text" - shows packet loss % / min/max, etc instead of only showing all the raw data.
- Can do some annotation of IP address in the history window so you can recognize what an IP was for (ie: enter an address of "10.0.0.1 My_Router" will auto-resolve 10.0.0.1 to "My_Router" even if not in a hosts table).
- Route change exclusion masks can be automatically calculated based on history. Makes it easy to exclude route changes for regularly oscillating routers at a hop (right-click on hop - does 255.255.XXX.XXX masking only).
- Final destination time-graph is *always* shown (unless explicitly supressed) - even if it doesn't respond (or gets lost from the route for some reason).
- Major re-write of alert system. Can now notify when a condition starts, when it ends, or as it is happening (as it is happening was only option before). Added the ability to launch an EXE when a condition occurs, and can also change the tray icon or show a message in the tray.
- Now supports Windows XP display schemes.
- DNS lookups use new Windows XP functionality when available to speed lookups (especially IPs with no names).
- Enabled history scrollbar on the "Address to trace" history when tracing.
- Added ability to multi-select and delete from "Address to trace" history (by multi-selecting and then right-clicking to pop up menu).
- Low latency hops (ie: less than 5 ms) would only average to whole numbers on time-line graphs. Now averages to accuracy of pixel size.
- Double-clicking on a hop number now shows/hides the time graph for that hop.
- Now defaults to "Best Method" of pinging. Uses high performance timers on OSes that need it (ie: NT/2000), standard method on others (ie: 95/98/ME/XP).
- Updated "WhoIs" parsing logic to redirect to the right registrar. Now defaults to whois.crsnic.net instead of whois.networksolutions.com for name lookups.
- Changed "Add Monitor" and "Modify Monitor" options to single "Watch this host (Alerts)..." menu to reduce confusion on setting up alerts.
Misc. changed functionality / bug fixes
- Fixed a bug in route change logic where data would sometimes get lost.
- Whois window greys the "Query" button while query is in effect (previous versions didn't give positive indication of current status).
- Route length oscillation no longer introduces incorrect packet loss in final destination.
- Re-worked logic in the way the "Route Change" list is populated so that frequent changes take a LOT less CPU time (previously, CPU could be at 100% - now at < 1% for same situation).
- Increased display speed on time graphs with big datasets (our test PC regularly runs at the 2 million sample mark - scrolling through old data is *WAY* faster (100X for some operations)).
- Splitters sometimes got lost - this is fixed.
- Clicking on a route now changes display immediately rather than having to triple or quadruple click sometimes (old bug).
- Can do Raw Socket ping instead of going through ICMP.DLL. (Note that this capability has no advantages in *most* situations).
- Moved all image logic internally and got rid of LPNG.DLL.
- Time graphs don't flash nearly as much - lots of flashing used to happen on route changes or as you scroll.
- Prompt for save on close now shows program name and destination (previously, lack of program name sometimes added confusion on shutdown of Windows).
- Fixed access violation when invalid hex string entered in packet cargo type / data.
- Disconnecting network card under Windows 98 sometimes didn't show timeouts. Fixed.
- Time graph rounding and cutoff were improved for smoother operation.
- Text Export shouldn't ever export a negative number now.
- (blatant teaser) Added basic support for new functionality in upcoming "Pro" version.
We did more to PingPlotter, even before this, but to see that history you'll need to visit our online release notes. If you think we should have all of history in this file, shoot us an email at support@pingplotter.com and let us know.