DBCC TEXTALL And SP5a (CRASH-ola!!!!)

Hi all,
I recently upgraded to SQL 6.5- sp5a. I run DBCC TEXTALL as part of my nightly maintenance. Our db is about 3 GB and is heavy on text
fields. Since the upgrade, DBCC TEXTALL hangs up the server until I reboot it. The last message in the error log is "Warning:
out of RA slots... comfigure more RA slots." (or something like that). This never happened before.
BOL says don't monkey with the RA slots per thread, so
I've been adjusting the number of RA worker threads. I raise it, bounce SQL SERVER, and try to run the DBCC TEXTALL again.
Same results. Eventually I got up to 10 RA worker threads, at which point I got an "IO buf_wait" error, which also hung up the server.
I've read that increasing RA worker threads will eventually run you out of buffer space, and it looks like that's what happened.

I'm beginning to think that the problem isn't really with isn't really with the number of RA worker threads I have configured,
but rather with something that sp5a changed. Or maybe DBCC TEXTALL is running into some sort of corruption it can't handle...
Or something...
So, I'm reading up on the relationship between buffer cache, RA, etc., but
the sad fact is that information on that low down level is hard to find, and I don't understand most of it anyway.
Any suggestions, reading recommendations, comments, or jeers would be highly appreciated.