Downloaded file not full-size. Why?

Download completes, but file not full size.

[correro]

I provide download on my website. When user clicks on Download button, most often download seems to complete successfully, but the file downloaded is not full-size, thus corrupt.

I tried various ways of coding on the client end, putting the URI in a hyperlink, or using a servlet to control download. None work reliably.

I'm getting confused, and desperate. What is the cause? The server (Linux+Apache+Tomcat)? The browser (I.E., Firefox)? File format (.exe, .zip, .rar)? Internet connection?

Can someone give me a clue? Thanks!

[MatthewHSE]

Hi there, and welcome to WebmasterWorld!

Is this an error that you're able to replicate? Or, does it seem to be happening only to a few users? Or, has it only been happening to you? Have you been able to pinpoint a common denominator, as in browser, OS, etc.? Also, does the download always stop at exactly the same file size every time, or does the actual downloaded file size vary at all?

[Farix]

Do you allow the use of download manager, such as GetRight or Download Accelerator? I've ran into several websites where I have trouble downloading a file--particularly large files--because the server drops the connection during the download. A good download manager has the ability to reestablish the connection and continue the download.

If the problem is persistent, then you probably should look into why the connection is getting dropped in the first place.

[Leosghost]

The latest firefox 1.06 seems to do this ..dont know about IE ( I wont let it connect anywhere out side of the machine it's on ) ..the problem is worse if you or your visitors installed the freedownloadmanager extension ( remove it )..the problem usually goes ..
...the problem is worse on zip than on other formats ..seems that it guages the total file size and then on certain zips ( can be as small as 500kb )convinces itself that it go all of it after only a few percent .."sumcheck" error? ...one work around is to make the file size visible in text at the point of the download click ..and if the user finds that they have smaller size than that at the "end" of their donload that they try to click "resume" ( tools :options :downloads: resume ) ..they may need to do "multiple resumes" or restarts to get the same file completed ..

Getright appears to have more "conflicts" when running under auto launch in firefox than others download progs ..

Another solution is to explain to them that they should... rightclick the item :copy link location ...then paste into the window of their download program ( normally the icon will be sitting waiting in their quick launch tray or toolbar tray )...

You might consider developing a revenue stream for yourself by becoming an affiliate to some of the better download managers..I include getright in those ( it just has problem with that "manager" plugin in Firefox )...place the aff links and some text explaining how great / cute etc such apps are to use on a page , and your download links along with 'em together with "if you already have brand X, Y or Z go straight ahead and download " if not "why not visit"...( I think they are all TBYB ) and make some aff money ...Or run adsense on the page and providing your text is well written and relevant to the products ..G will place the ads for you ..either way you can make some money out of a problem situation which may well not be of your own making ;)

[correro]

Thanks for your replies, guys. Here're some more details about my problem.

File size: 5.83M
Coding in the webpage:
1) direct reference to the file: <a href="/path/to/file">
2) using a servlet: <form action="/DownloadServlet">
Description: Download starts, and then finishes, apparently successfully. No error message is thrown. But 3 out of 5 times the file downloaded is smaller than the actual size. Downloads seem to succeed at some periods of time, while fail at some other periods of time.

I did some investigation, and found that download results do not seem to be correlated with file formats, coding in the HTML page, or browser, although Firefox outperforms I.E. in stability and reliability.

It looks to me more like a server issue, as if the server sends an End-Of-File or End-Of-Download signal prematurely.

On the server side:
OS: Linux
Web server: Apache
JSP container: Tomcat

The web-hosting service provider tried increasing Apache's Timeout value, but it doesn't seem to make much difference.

What could it be?