False Returns With Javascript Index

[Jeff Stallard]

I can't make my files available to you unfortunately, but here's a summary of what happens:

1. Create files Joe, John, and Jim.
2. Index the files, then publish everything to the web server.
3. Search for J. Click on search result Joe and the page displays as it should.
4. Delete file Joe from local machine and web server.
5. Index the files (so it's just John and Jim at this point), then publish everything.
6. Search for J. Joe still appears as a search result.
7. Click on Joe, and get a Page Cannot be Displayed error.

What I don't get is why Joe was included in the second index, after I deleted it. I ever looked at zoom_index.js and Joe doesn't appear anywhere in the file. Have you seen this before?

[David]

Maybe you are indexing in Spider mode and the file was cached? You can turn off caching from the "General" tab of the Zoom configuration window.

[Ray]

You also mentioned "zoom_index.js" which means you're using the Javascript version. As Javascript is run entirely on the client-side, this means that you might be testing your searches on a cached version of the search script. This would explain why "joe" doesn't appear in the index file but yet you are able to search for it. You will need to make sure you reload the search page properly (CTRL+F5 or CTRL+SHIFT+R in Internet Explorer).

[Jeff Stallard]

I tried everything you said, and no dice. I also deleted all cookies, browser history, and temporary files. Do you know the name and location of the cached index file (searched for zoom_index.js with no results)? If I can find it, I can verify that it contains the erroneous entries, manually delete it, and put this behind me.

[David]

It is quite impossible for Zoom to index a file that doesn't exist. So it is either a caching issue (move the files to another machine if you want to check this), or some silly error (like the word Joe appearing in one of the remaining files).

But if you can't make the files available to us, then it is pretty hard to really investigate the issue further.

[Jeff Stallard]

I finally found the problem. Previously, I verified that zoom_index.js did not include references to the deleted file. However, it turns out that zoom_pageinfo.js did contain the erroneous reference. I deleted all Zoom files and re-indexed, and everything works fine now. For the record, I typically do not delete the search files; I let Zoom overwrite them.

The time stamp for zoom_pageinfo.js was consistent with the other Zoom output files, so I know the file was being overwritten with each indexing (read-only was my first thought). I'm pretty confused how a fresh index could contain information about a file that no longer exists. By the way, there was also a file called zoom_pageinfo.tmp, which I deleted along with the other index files. Could this TMP file be part of the problem?

I'm going to watch things very carefully during the next few builds, and if it shows up again, I'll try to provide more detailed information about what happened.

[Ray]

That's strange. But it may be a possible consequence, for example, if you were holding one of the index files opened in another application (such as a text editor) while Zoom is attempting to update/overwrite them. Zoom should report an error in such case however and alert you to this. The TMP files are created during the indexing process, and used to produce the final index files, and immediately deleted after this. If they are still around, then I would suspect something like the above scenario may have happened, where a file was held opened for one reason or another, and Zoom was unable to write to the files it needed to. As mentioned though, I believe there should be errors reported in the Zoom logs when this happens, so let us know if it happens again.

(In general, don't open the index files until after Zoom has finished indexing them, and make sure no other application is trying to access the same files)

[Jeff Stallard]

That makes sense. I'll be more careful from now on. Thanks for the help.