Let me begin me begin by stating that my internet activity is a hobby -- not a business. It may be an addiction but it is still merely a hobby. It may be compulsive -- but it is a hobby. My wife has reminded me more than once that I have not made a penny off of the Web.
My local ISP has a number of classes of service, the most popular being the "Packrat" account which permits 600 hours per quarter and includes PPP and shell access as well as 10 megabytes of free storage online for web pages and other purposes for a reasonable fixed fee. The ISP also has commercial classes of service available for significantly higher prices.
The ISP updates the html.log for account in my class of service once a night, usually between midnight and 4:00 a.m. After that I run Analog2.11 to prepare my stats pages.
During a recent week the reports were truncated at mid day on two occasions and no data at all was provided for a third. After a few complaints appeared on in the support newsgroup, the owner of the ISP, responded somewhat at testily that if we relied on those reports we should move up to a commercial class of service, which makes hit data available immediately.
I do host some web pages for not-for -profit organizations. But my work on those pages is totally uncompensated and voluntary. The vast amounts of time I spend on those pages and other activities, my wife reminds me, could be used in changing light bulbs, mowing the lawn or-- God forbid -- sleeping!! And that time is not even tax deductible.
Like other hobbyists, I want the best. That is why I chose my provider. and why I stay.
My interest in my html.log is purely one of curiosity. It is fascinating to see what people are looking at and where they come from. It is interesting to find out if efforts to publicize my pages are successful. But even though I am just a hobbyist, I still want the information I receive and post to be as accurate as possible.
Since the Packrat setup does not allow running domain lookups by a cron*, I run my logfile analyzer whenever I wake up late in the night. There just isn't time when I get up for work. I use WS_FTP to assure that the log has been completed some time before I begin and then run Analog2.11. It may be that the fact that I want to run the analyzer, explains why I wake up at about 4:00 almost every night. Sometimes I have to wait for the file to be complete before I run Analog. I may be up an hour waiting to finish and to review the log.
After all that, it is disappointing to find the results to be truncated or missing.
I understand that another level of service is available. I understand that there is no guarantee for complete logs on the Packrat account I keep trying to convince my wife we can afford (even if one of our two 12 year old cars just died and our daughter is starting college and even if I keep griping about TVs and lights left on all night -- while this PC stays on so I can logon at the touch of a key without waiting to boot).
I have also witnessed my ISPs dedication to quality service over the years.
That having been said: when I posted an article pointing out that the log is truncated, it was not to demand a level of service beyond what I am paying for. It is to find out what happened. On a couple of occasions, for example. by posting a notice I learned that I was running Analog too early, before the complete log had been created.
Now I check to assure the html.log is complete before I post a message -- BTW: The file, which is approaches 4 meg a month, must be gzipped regularly to save space. - - The last time I tried to view the whole file it transferred at almost 10 kbps Not bad on an old 33.6 modem.) My notices are intended to alert the provider that something correctable may be happening on their system. Maybe the cause can be determined. I also post to determine if others hare having the same problem.
When the html.log was incomplete three times within a week I became concerned.
Several hours after I posted the message upon which this essay was based, my html.log was updated -- in the middle of the day. The missing data seems to have been restored. Shortly after seeing this message, the owner of my ISP posted a detailed technical response explaining some of the problems and making it clear that he had not intended to come down on the hobbyists.Leonard Grossman
* A "cron" or "crontab" is set on the provider's machine and causes a script or program placed on the server to run automatically at a set time. It can be used to perform many different functions.
Stephen Turner, who wrote Analog, assures me that "cron" comes from the same root as chronometer, and does refer to time.
If I didn't want the domain lookups so I could tell where hits come from, I could simply "set a cron" to run my "updatelogs" script at 5:00 a.m. every day. However for security reasons, my provider doesn't permit domain lookups to be run automatically. Thus I must run the application manually. And thus the bleary eyes.
The logfile analyzer I use is very configurable.
Go to my statistics page to see some of the kinds of data it will report. A link there will lead you to a page from which you can learn more about the application and can download the file. You can also get information directly at: http://www.statslab.cam.ac.uk/~sret1/analog/. I should also note that I am still using a commercial statistics service from Dbasics since I got in on a very a advantageous price a year ago. But that is about to expire. As I get comfortable with Analog, I will have to reevaluate use of the Dbasics service.
One interesting feature of Analog is the ability to define what will be considered a hit or a page. Therefore, I asked the program not to include graphics in its report (it would by default -- grossly inflating the number of hits.) OTOH, I did require it to record downloads of *.zip files as hits thereby giving me a record of the number copies of a shareware editor (DiDa) that are downloaded from my page each day. (Should I be happy or sad to discover that about half of the visitors to my home page come there to grab that file. Over a thousand copies downloaded in June.)
I have much to learn about additional configuration possibilities with Analog. For example, I understand it can read gzipped files. Since, as I mentioned above, my html.log grows to about 4 meg a month this would be very useful. As it is, I have had to create an archive containing monthly reports instead of keeping a continuous history.
If you got this far, I hope you got a smile out of it..unless of course you recognize yourself ;-)
Note: As of 7/11 the problem continues to recur. To this have been added some gaps in http accessibility on the server. Grrrrrrrrrrr!
Copyright 1997 Leonard Grossman
Send your comments or questions to email@example.com
My essays regularly appear in slightly different form in WindoWatch Magazine Which contains a wealth of fascinating information.
Back to my home page The ModemJunkie Speaks.
Back to Reflections of a ModemJunkie The complete archive of Reflections going back to 1992.
Created with DiDa!7/6/97 6:26:25 PM