Bad sector in raw OnLine chunk?

What is the best way to find and mark a sector of a disk bad that is in a raw Informix file? On an RS6000, while running a 4GL program, INFORMIX-4GL Version 4.10.UD3, INFORMIX-OnLine Version 4.10.UE4, we received the following error > Following fatal error occured during eqjabld20: > Date: 07/30/1996 Time: 08:51:27 > Program error at eqjabld20.4gl, line number 687 > SQL statement error number -408 > Invalid message type received from the sqlexec process Error # -408: -408 Invalid message type received from the sqlexec process. This is an internal error that shows a problem in the communication between the database engine and the library functions that call it. Make sure that your program is at the same software level as the database engine in use. If so, note all the circumstances and contact Informix Technical Support. The line where the program crashed was a simple insert: INSERT INTO eq_desc_181 ( cnt_nbr, proj_nbr, equipment_cde, comp_cde, sub_comp_cde, eq_desc, line_no ) VALUES ( gv_cnt_nbr, gv_proj_nbr, w_rec.equipment_cde, white_space, white_space, lar_eq_desc[i].desc, i ) >From the turbo-log: 08:46:22 Logical Log 96 Complete 08:48:34 Checkpoint Completed 08:50:25 Logical Log 97 Complete 08:51:27 newmode: Invalid page type of 0x0 08:51:27 bfcheck: bad page: pg_addr 0 != bp->bf_pagenum 10e215, userp = 30001cdc, pid = 11636, uid = 203 08:51:27 buffer header: 08:51:27 30388074: 00000001 00000000 30001cdc 303809b4 ........ 0...08.. 08:51:27 30388084: 3037e018 30389874 30382074 00020000 07..08.t 08 t.... 08:51:27 30388094: 02080000 0010e215 30575000 30001cdc ........ 0WP.0... 08:51:27 303880a4: 00000000 00000000 3038c19c 04000000 ........ 08...... 08:51:27 page header: 08:51:27 30575000: 00000000 00000000 00010000 e4b82b40 ........ ......+@ 08:51:27 30575010: 00000000 00000000 ........ 08:51:27 slot table and stamp: 08:51:27 30575ffc: 00000000 .... 08:51:27 -- Fail Consistency Check -- bfput -- pid=11636 user=203 us=30001cdc 08:51:27 bfcheck: bad page: pg_addr 0 != bp->bf_pagenum 10e215, userp = 30001cdc, pid = 11636, uid = 203 08:51:27 buffer header: 08:51:27 30388074: 00000000 00000000 00000000 303809b4 ........ ....08.. 08:51:27 30388084: 3037e018 30387c74 30000388 00030000 07..08|t 0....... 08:51:27 30388094: 03090000 0010e215 30575000 30001cdc ........ 0WP.0... 08:51:27 303880a4: 00000000 00000000 3038c19c 04000000 ........ 08...... 08:51:27 page header: 08:51:27 30575000: 00000000 0033b257 00010000 e4b82b40 .....3.W ......+@ 08:51:27 30575010: 00000000 00000000 ........ 08:51:27 slot table and stamp: 08:51:27 30575ffc: 0033b257 .3.W 08:51:27 -- Fail Consistency Check -- dovrecord:bad data page -- pid=11636 user=203 us=30001cdc 08:51:27 ERROR: logundo(40) iserrno 105 us 0x30001cdc pid 11636 tx 0x30003ae4 loguniq 98 logpos 0x10929c 08:51:27 INFORMIX-OnLine Must ABORT Log Error 'sprback() - logundo() FAILED' us 0x30001cdc pid 11636 us_flags 0x101 tx 0x30003ae4 tx_flags 0x222403 tx_loguniq 98 tx_logpos 0x10929c 08:51:27 -- Online Aborting -- us=30001cdc, pid=11636, uid=203 08:51:27 INFORMIX-OnLine entering ABORT mode!!! 08:51:27 -- Online Aborting -- us=30001b8c, pid=4742, uid=666 Tue Jul 30 14:28:43 1996 14:28:43 INFORMIX-OnLine Initialized -- Shared Memory Initialized 14:28:43 Physical Recovery Started 14:28:45 Checkpoint Completed 14:28:45 Physical Recovery Complete: 528 Pages Restored 14:28:51 Rollforward of log record failed, iserrno = 126 14:28:51 Log Record: log = 98, pos = 10929c, type = 40, trans = 2 14:28:53 Checkpoint Completed 14:28:53 ERROR: logundo(40) iserrno 126 us 0x30001cdc pid 0 tx 0x30003ae4 loguniq 98 logpos 0x10929c 14:28:53 INFORMIX-OnLine Must ABORT Log Error 'rollback() - logundo() FAILED' us 0x30001cdc pid 0 us_flags 0x121 tx 0x30003ae4 tx_flags 0x202403 tx_loguniq 98 tx_logpos 0x10929c 14:28:53 -- Online Aborting -- us=30001cdc, pid=4742, uid=666 So we may have a bad section of the hard disk? 08:51:27 bfcheck: bad page: pg_addr 0 != bp->bf_pagenum 10e215, userp = 30001cdc, pid = 11636, uid = 203 Maybe the bad section is where a log was written? 14:28:51 Rollforward of log record failed, iserrno = 126 Trying to bring informix back up using tbmonitor gives the error that the OnLine daemon is no longer running. The log listing (tbstat -l) showed that about half the logs are full. Informix cleared the logs with tbzero but the identical problem occured again later, also during a write. _____________________________________________________________________ | Colin McGrath cmm@trac3000.ueci.com | | Raytheon Engineers & Constructors, Inc. (215) 422-4144 | | Phila, PA, USA Standard Disclaimers Apply | |___________________________________________________________________|