"Al VAs" <deleteprefix_alex@favour.com.au> wrote in message
news:247566A4-023D-41C2-8DD7-7A92314024A9@news.elevatesoft.com...
> Hi,
>
> Does anyone use pessimistic locking on their tables?   We do, very
> important for this table, maybe that is an issue?

Optimistic here.

R

Hi Roy,

Been reading and found somewhere it saying that performing a flushbuffers on
idle would be a good practise. Would this be a good idea?  would mean no
extensive slowness as it is only occurring on idle application.

Thanks

Alex
"Roy Lambert" <roy.lambert@skynet.co.uk> wrote in message
news:6056C13B-B8FF-4ECE-AB8E-7B4FAE2CF0B4@news.elevatesoft.com...
> Al
>
>
> What happens if you add a flushbuffers after each post. I know it will
> slow things down but it might make things work.
>
> Also, this is going from memory, and probably not relevant, when I was
> running TMaN under V3 I had to reduce the number of background threads
> down to 1 from 4.
>
> Memory says that was pre 3.30 and it didn't exactly "cure" the problem
> just meant that things were running slow enough the system could keep up.
>
> Roy Lambert
>

Alex,

<< Been reading and found somewhere it saying that performing a flushbuffers
on idle would be a good practise. Would this be a good idea?  would mean no
extensive slowness as it is only occurring on idle application. >>

Unfortunately, FlushBuffers won't have any bearing on this issue.

--
Tim Young
Elevate Software
www.elevatesoft.com

"Tim Young [Elevate Software]" <timyoung@elevatesoft.com> wrote in message
news:753FA6C7-D4E5-4EF2-9CE2-66DAD2717AE6@news.elevatesoft.com...
>
> Unfortunately, FlushBuffers won't have any bearing on this issue.
>

If you didn't insist on sleeping 4 hours a night, we would already have a
solution.

Robert

Al


It might help, but at least in your case I doubt it. The OS is meant to flush its buffers every so often anyway so its unlikely that just flushing them on idle would cure the problem but who knows.

In my case we finally found what was causing the problem (as Tim indicated it took a looooong time but eventually I came up with a demo app that caused the problem every time) but one of my early attempts simply slowed things down a bit so that DBISAM didn't get the index buffers out of sync.

It might be worth while having a user settable flag to do a flush buffers aster each post on the problem table. If that works on the affected sites at least you have an interim solution albeit not a good one.


Roy Lambert

Robert,

<< If you didn't insist on sleeping 4 hours a night, we would already have a
solution. >>

I'm actually getting 6-7, which is pretty good considering I can't get my
mind off of the EDB stuff that I've got left.

--
Tim Young
Elevate Software
www.elevatesoft.com