As noted previously, the fandle busy problem that I described in the past reoccured even after the system was patched.
I have just had confirmation from HP Engineering that a further timing window has been discovered that could also produce the error.
So, the original patch wasn't dropped. It just didn't cover all eventualities.
Posted at October 6, 2005 1:19 PMyour previous entry on this has several incorrect links.
I've not seen many programs using fast I/O and am interested to see if it makes a big difference.
Posted by: Ian Miller at October 11, 2005 12:30 PM
Ian, thanks for pointing out the link weirdness. I suspect that was a side effect of moving to a new web host recently. I've corrected the links and gone back through the entire site looking for others too.
As to how much difference fast I/O actually makes, I don't know, although I expect it's significant. Fast I/O is using preallocated buffers from the BAP, and so doesn't have to do page locking and so forth.
Posted by: Jim Duff at October 11, 2005 8:21 PM
Comments are closed