I have ~26,000 documents in the database which are very similar (e.g.,
containing attributes) to the problem documents, and I don't see this
behavior until I insert a document with the 4K attribute.
BTW, I should be more specific: the attribute is the CONTENT attribute
of the META tag.
> You will get this even if you have no attributes. All that is happening is
> that mssearch is consuming all available cpu. Higher priority
> services/processes can but in and take cpu cycles away from mssearch/mssdmn.
> What you should do is stop mssearch and watch the available Mbytes note the
> peak and then start up mssearch again. Wait a day. Stop mssearch again,
> see if the available Mbytes returns to the same value. If available bytes
> keeps on getting lower and lower you know you have a leak. If it keeps on
> returning to the same value (or thereabouts) there is no memory leak in
> mssearch or mssdmn.
> Another clue that you have a memory leak is you server will blue screen at
> predictable intervals - for instance every Thursday around noon.
> > Hi all,
> > I'm indexing an image column with an associated file type column on
> > SQL 2000 SP2 on a dual processor server with 1GB of memory. The
> > content is all HTML. I have found that if the HTML contains an
> > attribute of 4095 characters or more, mssearch and msssdmn start using
> > lots of CPU time (eventually reaching and sustaining 100%) and
> > consuming all the available physical memory.
> > The wisdom of a 4K+ attribute value is certainly questionable, but I
> > understand that HTML 4.0 allows for 64K. At any rate, I encountered
> > this problem with the actual data that I have to work with, so if
> > there isn't a fix out there, I'll have to filter and truncate 4K+
> > attributes before inserting them into the database.
> > Has anyone else encountered this problem? I have not been able to find
> > any bug reports about it.
> > Thanks,
> > Dave Lemen