Subject Re: [firebird-support] Re: Firebird Restarting part 2
Author Vlad Horsun
""Michael Vilhelmsen"" ...
> > Well, for whatever reason you're flat out of memory. Here's where that
> > error is generated in cch.cpp, CCH_init.
> >
> > /* Log if requested number of page buffers could not be allocated. */
> >
> > if (count != (SLONG) bcb_->bcb_count)
> > gds__log
> > ("Database: %s\n\tAllocated %ld page buffers of %ld requested",
> > tdbb->tdbb_attachment->att_filename->str_data, bcb_->bcb_count,
> > count);
> >
> > The call above to memory_init calls gds_alloc which calls malloc and
> > the malloc failed to get the whole page cache. The question is why.
> >
> > Your system has lots of memory - 2Gb and you'r asking for about 8 Mb.
> > What else is running on that system?
> >
>
>
>
> My brother (who keeps the serves up to date) has produced some
> pictures of whatever the server is doing.
>
> They can be found here:
>
> http://download.microcom.dk/firebird/fb1.jpg
> http://download.microcom.dk/firebird/fb2.jpg
> http://download.microcom.dk/firebird/fb3.jpg
> http://download.microcom.dk/firebird/fb4.jpg
> http://download.microcom.dk/firebird/fb5.jpg
>
>
> The last 2 pictures shows, that the usage of the pagefile is some 55-60%.

Look at last picture - fbserver have Virtual SIze = 1.9 GB but used only
0.58 GB physical memory (working set). I.e. requests to allocate more virtual
memory will fail (1.9GB very close to limit of 2GB per process) despite of free
memory available.

> This server is dedicated to run firebird and nothing else.

How many databases active at this moment ? Does some of them have
fixed cache setting ?

> At this time we suspect that the users actually are doing some rather
> heavy reports.

Does you use some custom UDF (with memory leakage) ?

Regards,
Vlad