Problem with auto save data

Posted by: kalm

Problem with auto save data - 03/31/00 06:22 AM

Hi,<br><br>I have set up autosave on the two instances of PingPlotter that I'm runing.<br><br>ver: 2.20 beta 1 running on Win98<br><br>Auto-Save data is checked<br>Save interval 30 minutes<br>filename: $host.$date.$hour.$minute<br><br>And the filename in the tool tips box looks allright to me.<br><br>Still no data is saved.<br><br><br>
Posted by: Pete Ness

Re: Problem with auto save data - 03/31/00 09:43 AM

I had this happen to me when I had a colon (:) in the file name. Also, had it happen when I had a slash in the name. Any chance you may have a : instead of a .. anyplace?<br><br>How about using a set file name (no variables)?<br><br>Thanks the the report (and help isolating!).<br><br>Pete<br><br>
Posted by: kalm

Re: Problem with auto save data - 04/01/00 09:19 AM

Hi,<br><br>I didn't have tive to to a closer examination, short on time.<br><br>It might have to do with a filename length limitation...<br><br>I had the filename set to: $host.$date.$hour.$minute and this didn't work, changed it to<br>de.$date.$hour.$minute and the problems wen't away.<br><br>I was tracing, so that would result in a quite long name.<br><br>I will have more time to test on Monday.<br><br>Kim <br><br>
Posted by: kalm

Re: Problem with auto save data - 04/03/00 04:32 AM

Hi,<br><br>Tried to replicate the problem with auto save, but I wasn't able to have the same problems as I had on friday, and filename length doesn't seem to be a problem.<br><br>But I came up with a few issues.<br><br>Setting the save time to less than a minute is not very usefull, it will overwrite the previous save... not a major problem.<br><br>Setting the filename to $host\$date.$hour.$minute works great, this saves all data for the pinged host in a directory of it's own. But no save will happen if the directory does not exist, that's not a problem for me, because the number of host's that I want to keep track of is quite stable.<br><br>The last item on this topic, the auto save does save the entire sample set, and I have the number of packets to keep in memory set to 5000, one ping every 5 second, and autosave set to 30 minutes, so this gives me a lot of overlaps, so I would prefer the autosave to save only the data in the autosave period.<br><br>So instead of saving all 5000 packets on every save, it would only save the 360 collected during the half hour interval.<br><br>Kim<br><br>
Posted by: Pete Ness

Re: Problem with auto save data - 04/03/00 12:24 PM

Hi Kim.<br><br>Thanks for the followup and feedback. I've got some questions for you - and some comments. I'd like to make sure I'm solving the right problems - and coming from the right direction.<br><br>First off, overwriting previous saves is something I intended to use to make sure data is saved. A few people had expressed concern about making sure their data is saved - and the save interval reflects the ability to save as often as you want to. <br><br>Different save intervals can actually be used with any filename. For example, in my testing, I found that I liked to have save sets separated by hour, but I saved much more often.<br><br>In your case, for example, you might set up your filename to be: to $host\$date.$hour.00. You could then set your save interval to something pretty low (say 5 minutes), and you'd save every 5 minutes, but only the last save in an hour would "stick". Any previous saves would be overwritten, and would happen mostly just to make sure you got a save (in case you rebooted, lost power, whatever).<br><br>The "Auto-Save Data" and "Auto-Save Image" really evolved a bit together - and using this kind of mechanism with "Auto-Save Image" works great - save often, but use a common filename. With images, you might just use "$host", and then just overwrite the file a *lot* (once a minute?). Doing that, you would have a fixed name that you could include on an intranet page (or, possibly, internet, if you put something in place to get that image to a publishable site).<br><br>Ok, with that background in place ...<br><br>Personally, I kind of like to have at least *some* overlap. This makes sure I can entirely see problem periods - and 30 minutes isn't really enough. With only 30 minutes saved in a file, you have a pretty good chance of having to load up a previous file. My vision of this would be to have a few overlapping files so I don't have to reload. <br><br>Of course, having 24 150K files every day may not be so good for you either - and there's no position between and It sounds like in your case, 4 save files a day would work pretty well (still saved every 30 minutes if you wanted, or more or less often). There's no way to get ($hour div 6), though.<br><br>So, there's a few possible angles I can see here.<br><br>1) Add a "time period" to save. If you're saving every 30 minutes, have another parameter that says how much data to save (in time, rather than sample counts).<br><br>2) Allow for some way of saving fewer files than .day.hour, but more than .day.<br><br>I will add the capability of having it create the directory if it needs to. I agree that the $host\$date... thing could be pretty neat.<br><br>I'm interested in hearing your (and everyone else's) thoughts on this.<br><br>Thanks!<br><br>
Posted by: Pete Ness

Re: Problem with auto save data - 04/21/00 12:36 AM

Ok. For beta 2, it will auto-create the directories. None of the other things I talked about here are implemented, though (we'll see about getting them in for the *next* release).<br><br>
Posted by: Kenric

Re: Problem with auto save data - 03/17/01 08:19 AM

How about an option to only save whatever amount is indicated by the auto-save filename, and in addition to that (now here's my cool idea), add the ability to load multiple sample sets (hold down Ctrl while selecting files in the Open dialog) as long as they are for the same host.<br><br>For example, if you set your auto-save filename to:<br><br>$host $date<br><br>By the end of each day you'd have a full day's worth of samples in each file, with no overlap (solves the problem of files getting too big over time), and you wouldn't need overlap if you can load multiple files to view as many consecutive days worth of samples as you need.<br><br>-Kenric<br><br><br>