--- In firebird-support@yahoogroups.com, "Ann W. Harrison"
<aharrison@...> wrote:

>
> Vlad has done some interesting examination of the
> index handling code and offered a second possible
> explanation of these errors.
>
> My theory requires that the server or at least one
> of server processes crash. Without crashes, my
> theoretical problem can't occur.

Well, my server processes do crash at the time but I'm pretty sure the
crash is a result of the corruption and not the cause of it. I have
looked back through the firebird logs and found the following entries
in all three of the corruption events I have seen.

****************************************************************
KSSHEFAX01 Fri Dec 01 09:44:30 2006
SERVER/process_packet: broken port, server exiting

KSSHEFAX01 Fri Dec 01 09:50:40 2006
Database: E:\CALLSYS\LEEDS\TEST\DB\CALLSYSLEEDSTEST.FDB
Index 4 has orphan child page at page 9933 in table ACTIONS (171)


KSSHEFAX01 Fri Dec 01 09:50:43 2006
Database: E:\CALLSYS\LEEDS\TEST\DB\CALLSYSLEEDSTEST.FDB
Index 19 is corrupt on page 11090 in table ACTIONS (171)
****************************************************************


****************************************************************
KSSHEPRFX (Client) Wed Nov 22 12:10:44 2006
INET/inet_error: send errno = 10054

KSSHEPRFX (Client) Wed Nov 22 12:10:44 2006
REMOTE INTERFACE/gds__detach: Unsuccesful detach from database.
Uncommitted work may have been lost

KSSHEPRFX (Client) Wed Nov 22 12:10:44 2006
INET/inet_error: send errno = 10054

KSSHEPRFX Tue Nov 28 12:25:54 2006
Database: F:\CALLSYS\SHEFFIELD\TEST\DB\CALLSYSSHEFTEST.FDB
Index 6 is corrupt on page 111025 in table ACTIONS (164)


KSSHEPRFX Tue Nov 28 12:25:54 2006
Database: F:\CALLSYS\SHEFFIELD\TEST\DB\CALLSYSSHEFTEST.FDB
Index 6 is corrupt on page 111025 in table ACTIONS (164)
******************************************************************

******************************************************************
KSSHEPRFX Wed Dec 06 23:12:24 2006
SERVER/process_packet: broken port, server exiting

KSSHEPRFX Sat Dec 09 09:32:58 2006
INET/inet_error: read errno = 10054

KSSHEPRFX Sat Dec 09 09:34:24 2006
Database: F:\CALLSYS\SHEFFIELD\LIVE\DB\CALLSYSSHEFLIVE.FDB
Index 6 is corrupt (missing entries) in table ALLOCATION_HISTORY (155)


KSSHEPRFX Sat Dec 09 09:34:25 2006
Database: F:\CALLSYS\SHEFFIELD\LIVE\DB\CALLSYSSHEFLIVE.FDB
Index 7 is corrupt (missing entries) in table ALLOCATION_HISTORY (155)

*******************************************************************

>
> Vlad's theory requires that entries be removed and
> inserted in the same value range to the extend that
> an index page first splits and is then recombined.

I never delete from the tables that are affected by the corruption.
It is concievable that a rollback could get called on a failed insert
though. if this did happen the application would try the insert again.

I do however perform frequent updates on foreign key/indexed fields
which may change the field to one value and then back again. Would
this do the same as a delete/insert in the index?

>
>
> So, for those of you who have highly concurrent
> applications running 1.5.3 Classic and are seeing
> occasional corrupt indexes with a wrong page type
> in the index (expected 7 encountered 5) - do your
> environments resemble either of these?

Ann, I should point out that I have not got any evidence in the logs
that I have, that any of the corruptions I have seen are (expected 7
encountered 5) types. You pointed out that I am probably seeing the
same issue as the other folks here.
I just wanted to be clear about that as I didn't want to send you on
a wild goose chase!

Regards,
Will.