Losing connectivity

If I run PingPlotter Version 2.20 Beta 1 for more than a couple of hours pinging my cable modem gateway, I lose the ability to access my cable modem DNS servers. I am connected using DHCP on a Win98 SE system. To fix the problem, I must reboot my system.<br><br>I supply specific IP addresses, if requested.<br><br>

Hmmm.<br><br>This is a new one on me.<br><br>Does version 2.03 do anything similar? <br><br>If you ping beyond your gateway (ie: a web site or any IP beyond your gateway) does the same thing happen?<br><br>Do you have any other machines hooked up through your cable modem? (I'd be interested to know if only one machine is unable to access it, or if it's network-wide).<br><br>Let me know any more info you have here...<br><br>

I discovered that the 2.20 Beta 1 version has a tremendous memory leak. If left to run for several hours, it eats up so much memory that many functions, including my browser, fail to work properly. Attempts to load new programs result in "out of memory" messages when I have 256MB of RAM.<br><br>In answer to your questions:<br>1) I did not have this problem with 2.03.<br>2) I am only pinging the first hop in my network.<br>3) I do have a second machine attached, but problems with my ISP (the reason for running PingPlotter) preclude me knowing if the problem extends to my other computer.<br><br>I am running Win98 SE on a dual Celeron 466 system with 256MB of RAM.<br><br>

Hmmmm. Can you possibly check the date/time on your executable? I fixed a huge memory leak right before I released 2.2 beta 1. My win98se test machine has been running for about a week solid and has collected about 120K samples. I'm wondering if it's possible that you *somehow* got an old version, or if this is a new memory leak that only happens at certain times (ie: 1 hop long routes? < 30 ms response times? Non-responding DNS servers? Any other ideas?).<br><br>My .exe is dated 3/15/2000 - and is 799232 bytes. I'd love to catch any bug with this before the next beta (which is getting pretty close).<br><br>Thanks!<br><br>

Thanks for this info, Reg. Got it tracked down. <br><br>Summary: Any time a host is unreachable, it's creating a new temporary "host" placeholder for every request that's sent out for a hop after the last router that returns any data. Over the course of some time, this can be a substantial number - and each is filled with data up to the current sample (which is pretty dang painful in memory usage...). <br><br>This will be fixed in beta 2.<br><br><br>