> We have a server running Fireird 1.5.5.
> Some 20 DBs are active at the same time.
> There are between 1 and 100 users connected to each DB at all time.
> The DBs varies from 20 Mb to 3 Gb.
>
> Every once in a while the Firebird server crashes.
> This gives on of 2 senarios.
>
> 1.
> The guardian takes over, restarting the server.
> Everybode gets an "Unable to connect to database..." (or something
> like this)
>
>
> 2.
> No one can connect to the DB, and those connected cannot do anything
> without getting every kind of error.
> Stopping and restarting the Firebird server solves this.
>
>
> Everytime these situations happens we can see that almost all of the 2
> Gbs of installed RAM are used.
>
> Usually we get this entry in the log file:
>
> DBSRV05 (Server) Wed May 14 12:39:03 2008
> Database: E:\DB\MISSYASKIVE\OCCLICENS.FDB
> internal gds software consistency check (cannot start thread)
>
>
> But just recently we have starting to get this error as well:
>
>
> DBSRV05 (Server) Wed May 14 12:38:59 2008
> Database: E:\DB\MISSYASKIVE\OCCLICENS.FDB
> Allocated 813 page buffers of 2048 requested
>
>
> The last one is a new one.
> And until we stop and start the Firebird server it will stay like this
> forever.
>
> Question is - Those this mean, that Firebird are requesting more page
> buffers than the OS can give ?

This mean that Firebird's process already allocated 2GB of VM and
failed to get more.

Usually this is because of some high value of "DefaultDbCachePages"
setting in firebird.conf or database(s) have high value of "page buffers"
configured at header page.


Regards,
Vlad