Topic Options
#225 - 07/15/00 01:20 PM Values in log file are truncated
Marc_Kupper Offline

Registered: 07/11/00
Posts: 12
Loc: California
I'm running 2.2b5 with an Alert set to log if a single ping is over 1000ms. (I want to log all slow pings and failures). My alert is attached to all hops of the route. I'm running a ping every 15 seconds with unlimited/all.<br><br>I am seeing things like the following in the log file.<br>Date/Time 1 2 3 4 <br>7/15/2000 12:56:56 AM 17 11 16 21 <br>7/15/2000 1:37:11 AM 31 20 17 24 <br>7/15/2000 2:54:12 AM 25 20 15 19 <br>7/15/2000 4:15:57 AM 35 10 15 20 <br><br>None of the numbers are 1000ms or longer though that's what my altert is set up to log.<br><br>I'm not sure if this is related but my present time-line shows a single red line right at about the 1:37am mark. What's add is that the red bar does not extend to the top of the time-line line as with a normal failure. There is a log entry for 1:37am. Some time during the night the Err count did jump for all hops but no errors were logged.<br><br>A few days ago there was a hard failure and it was logged as<br>7/11/2000 8:43:44 AM -327 10 20 40 <br><br>Marc<br><br><br>

#226 - 07/17/00 12:18 AM Re: Values in log file are truncated [Re: Marc_Kupper]
Pete Ness Offline

Registered: 08/30/99
Posts: 1106
Loc: Boise, Idaho
Interesting!<br><br>Ping Plotter attempts to give you a "picture" of the number of samples around the error condition that you say you care about. It uses the "Traces to examine:" number in your alert and packages up this number of samples into the email or text file so you can see if it's an isolated case or a trend. Looks like possibly if this is set to 1, then it grabs the one before or after the error condition. Probably, increasing the number of traces to examine would give you a temporary workaround for this problem.<br><br>The red lines aren't full height if a pixel represents more than one sample - and more than one sample in that time period was good. It averages the good samples in the period (just like always) but paints it in red to show that there was at least one lost packet in that period.<br><br>The -327 you saw is a bug in the output logic - I changed the timeout constant to be -32767 instead of 9999 - and I missed one area of fixing that. Thanks for the report.<br><br>I'll have a look at this logic to make sure it's working correctly. Thanks for the bug report.<br><br><br>



Who's Online
0 registered (), 13 Guests and 1 Spider online.
Key: Admin, Global Mod, Mod