Keep getting Access violation at 0x7C918FEA

[sdmsearch]

hi

why do i keep getting this error all the time i index

[David]

That error with the crash address, 0x7C918FEA is probalby not from our software.

Thousands of people not using our software have the same error, see,
http://www.google.com/search?q=0x7c918FEA
I read about a dozen pages from people with the same error, but most of it seems to be mis-information, or people asking about this crash but getting no sensible answer. Some people are claiming it was
- mal-ware,
- others claimed it was 'HeapAgent' from MicroQuill
- still others claimed it was a anti-virus program causing the problem,

but it didn’t seem conclusive.

If you have any additional information, please post it, in case someone else has the same issue.

[Update: Turns out this was wrong. It was a memory corruption problem caused by a software bug in our code, see below. The other reports on the Net for this same crash address are proably also due to memory corruption, in other words a software bug in the respective products]

[Ray]

Can you confirm if the Zoom Indexer window stops responding, or disappears when you see this error message?

And from your previous posts, it would seem that the Zoom Indexer ran successfully on your machine in the past. Can you tell us what has changed on your system (e.g. if you have installed or updated any software) between then and now?

[sdmsearch]

i get this error

hi ray

the site i get the error on is http://news.sky.com i thnk its down to the robot.txt on the server will you have a look and ive not installed any software or windows updates i have update the ver of zoom search 5.1 1011 to zoom search 5.1 1012 thats it

thanks paul

[Ray]

Sorry Paul, I must have missed this message somehow. I've checked the problem since, and can confirm the crash when indexing that website with robots.txt enabled.

The problem is due to the robots.txt file on that website being extremely long and having over 900 disallow entries. Zoom is only supporting a maximum of 200 skip entries per start point, so this goes over the limit. It should never crash however, and we will need to fix this for the next public build (5.1.1013).

Just so you know though, this problem would have existed in builds prior to 5.1.1012, and it may just have been luck that you did not come across it before (or you may have robots.txt disabled, or you were not indexing this particular website).

We will consider increasing the limit from what we currently have, but it is worth noting that this 900+ disallow robots.txt file is quite unusual, in that it is mostly redundant. For example, it contains entries like the following:

Disallow: /customerzone/
Disallow: /customerzone/articlegrid
Disallow: /customerzone/article
Disallow: /customerzone/article/article
Disallow: /customerzone/betvegas
Disallow: /customerzone/blank1
Disallow: /customerzone/channelnav
Disallow: /customerzone/christmas
Disallow: /customerzone/common
Disallow: /customerzone/common/channelPageSnippet
Disallow: /customerzone/common/errormessage
Disallow: /customerzone/common/gridPageSnippet
Disallow: /customerzone/common/message
Disallow: /customerzone/common/sessioninfo
Disallow: /customerzone/community
Disallow: /customerzone/competition
[... and it goes on for several hundred]

When in fact, the first line of the above already disallows all subdirectories under "customerzone", so it is all fairly pointless, and a badly written "robots.txt" file. But we'll see what we can do. In the meantime, you might want to disable robots.txt support or remove this site from your index.

[Ray]

Actually, on closer inspection, build 5.1.1011 might not have exhibited the problem because it contained a bug where it was ignoring multiple disallow entries (and thus never exceeding the limit). But that would be little different to disabling robots.txt support in 5.1.1012 as a temporary workaround.

[Ray]

Update: This bug is now fixed in build 5.1.1013 which you can download from here.

[sdmsearch]

keep getting this erorr back

hi ray

would you have a look at this site im indexing as it's giving me the same error as the sky.com was the site is www.bbc.co.uk when index up to 3000 zoom search software locks up and gives the error the index was running just up to one day 22 hours

is the a way of saving the last index files so i don't have to keep starting over ?

thanks paul

[David]

Can you clarify what you mean by, "when index up to 3000 zoom search software". Did you mean 3000 pages from the BBC site?

Can you E-mail us your current configuration file and confirm what Zoom version & O/S you are now using?

Is there any reason you think this is related to the 0x7C918FEA crash (and not a different issue)?

[sdmsearch]

new error with zoom search 5.1 1013

hi ray

here is the error what i get from zoom search 5.1.1013


i dont know why it keeps doing this ? the web site its doing it on is www.itv.com and www.bbc.co.uk

thanks paul

[David]

I think this is a different issue from the orginal post.

Again there are a lot of people on the Internet claiming this 0x7c901010 crash address is related to viruses and malware in IE. But everyone blames malware for everything nowadays, so who knows.

There was some basic analysis done in this game development forum. It seems this 0x7c901010 crash is due to the calling the Windows operating system function EnterCriticalSection() with bad parameters. In other words a software bug. But the bug could might not be in our code (it could even be a bug in some malware code or it O/S itself).

I think Ray did some testing on this yesterday. So he'll probalby have something to add.

[Ray]

We could not reproduce the crash, using the ZCFG file you sent us. It indexed beyond the www.itv.com and www.bbc.co.uk sites without problem.

Have you tried (as a test), removing all the other start points, and only indexing these two sites? Does it still crash, and if so, at what point? You can turn on debug logging to potentially narrow it down to a particular file.

We also noticed you had very large limits specified in your ZCFG file (1 million pages and 1.1 million unique words). It does not seem like you are indexing up to this number given the URLs specified. It is best to set a limit that is really closer to what you need indexed, because the higher the limit, the more resources Zoom will use. If your computer is physically restricted (due to amount of RAM available, etc) then you may force Windows and other applications (including Zoom) to run in an unstable environment.

It is possible, for example, if you do not have much RAM, that this crash is occuring in an obscure situation where Zoom requests for more RAM but is denied by Windows. And that we did not reproduce this because we have more RAM available and the situation did not arise. How much RAM is installed on your computer?