INFORMIX SERVER / 5725A3900 / C10 - IDS 12.10

Table is created in a 4K page dbspace.  It has a 1 single data
page in use with 59 rows on it.  It is then altered by 7 alter
table statements that add 4 columns (1 int, and 3 varchars, and
then reduces the size of the 3 varchars that were just added).

At that point an update statement is run that modifies just 1
row on the page.  The server appears to think it can leave all
the rows on the same page, and ends up writing into the slot
table (because there isn't enough free space on the page to
accommodate the increased row size for all 59 rows).

The server assert fails trying to write the page out (so it's
caught in the bfcheck that bfput is doing) but the changes get
logged, so when the server crashes, it attempts to do the same
thing in recovery and fails again, so the server can't be
brought online.

The stack trace in the failure:

afstack
afhandler
affail_interface
bffail
buffput
ckpgversion  (may not see this frame, was present on solaris,
not present on linux)
rewrecord
rsrewrec
fmrewrec
aud_sqisrewrec
doupdate
chkrowcons
dodmlrow
dodelupd
excommand
sq_execute
sqmain

****************************************************************
* USERS AFFECTED:                                              *
* Users of IDS prior to 12.10.xC13.                            *
****************************************************************
* PROBLEM DESCRIPTION:                                         *
* Bfcheck bad page pg_frptr  < sizeof(ifx_page_t) 24 or  *
* > slotbeg  when updating page with outstanding in      *
* place alters.                                                *
****************************************************************
* RECOMMENDATION:                                              *
****************************************************************

keiner bekannt / siehe Local-Fix

Datum - Problem gemeldet    :
Datum - Problem geschlossen :
Datum - der letzten Änderung:

05.02.2019
24.09.2019
24.09.2019

12.10.xC13