Two days of downtime - could rollback have been used here?

Recently, our informix-online was down for more than two days. We have a 4 hour call service contract with our wendor. The problem gave this in online.log: 17:31:47 Checkpoint Completed 17:35:27 bfcheck: bad page: pg_flags 4 != type 1, userp = 40001444, pid = 1.4967 , uid = 5945 17:35:27 buffer header: 17:35:27 40135794: 00000000 00000000 00000000 40123b24 ........ ....@.;$ 17:35:27 401357a4: 40131454 4012a794 4013b394 00020000 @..T@... @....... 17:35:27 401357b4: 20030000 00106c04 4048b000 00000000 .....l. @H...... 17:35:27 401357c4: 00000000 00010000 4015319f 01000000 ........ @.1..... 17:35:27 page header: 17:35:27 4048b000: 00106c04 0032e41a 00000004 00180fe4 ..l..2.. ........ 17:35:27 4048b010: 00000000 00000000 ........ 17:35:27 slot table and stamp: 17:35:27 4048bffc: 0032e41a .2.. 17:35:27 -- Fail Consistency Check -- isread:bad data page -- pid= 1.4967 user=5945 us=40001444 17:36:48 Checkpoint Completed It somehow looks like a hardware error, but the disk is mirrored. The user was kicked out. Later he, and other users of the same database also were kicked out with an error. The vendor tried to export the relevant database, but it would not export. Though, it was possible to export the tables, one at a time. It ended up with that all databases were exported, the 1GB chunk re-initialized, and then the system seems OK. The disk was verified during this. I think a single soft error was found. Now the question: With Informix-online's nice roll-back feature, couldn't it have been fixed by restoring the backup tape of the whole chunk and doing a roll-forward? That would have been a lot faster. Please email. I do not receive this group regularly. -- tfj@olivia.ping.dk