Strange logs of PDF downloads

Why are there multiple requests for a single file?

[craig_d]

The download of a PDF from my server (Win2K v5 sp2 with IIS 5.0) creates multiple requests in my log files...see below. Is this a problem with server settings, the PDF, impatient users, or something else? Sometimes there are 20-30 separate entries for one file. Please help!

Here is the page request:
date time c-ip s-ip GET /bmr2000.html - 200 11950 USER-AGENT
Then the product image request:
date time c-ip s-ip GET /images/products/brushless_servo_motors/bmr2000.gif - 200 15737 USER-AGENT
Now the trouble begins:
date time c-ip s-ip GET /downloads/bmr2000.pdf - 200 0 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 0 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 50072 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 46034 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 1171 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 1291 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 1291 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 420 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 610 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 1816 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 419 USER-AGENT
date time c-ip s-ip GET /downloads/bmr2000.pdf - 206 126234 USER-AGENT

This particular file (bmr2000.pdf) is 370KB

A direct link to this page is:
<Sorry, no personal URLs. See TOS [webmasterworld.com]>

Thanks in advance for your help!

[edited by: tedster at 8:39 pm (utc) on Nov. 24, 2004]

[cgrantski]

It's because the pdfs are being broken into chunks to give the user a quicker view of the pdf, or the beginning part of it anyway. It's controlled at the client end. It doesn't indicate any kind of problem.

Because the first chunk has a 200 code and subsequent ones have 206 ("partial") codes, it's possible to track pdfs pretty easily by ignoring the 206's. It's also possible to get a decent idea of whether the entire document was downloaded by reporting on the 206's.

[craig_d]

Thanks for the info cgrantski.

I've tested this but there seems to be no measurable relationship between the number of pages in the PDF and how many pages are clicked through vs. the number of "hits" to the log file.

Could you give me a bit more detail on how I could know the number of entire downloads based upon 206's?

Thanks again.

[cgrantski]

I've never seen the relationship either. I think different browser/download manager combinations divide them up differently.

But try adding up the file sizes and comparing to the actual size. That's what I should have said more clearly. I don't think there's a perfect, to-the-byte match, but it's close enough.