phil@attctc.Dallas.TX.US (Phil Meyer) (09/30/89)
This is a horror story, as well as a sincere plea for help. First, the facts. There is no way to verify the backup tapes produced by informix turbo. You cannot use standard UNIX tools to play with the raw dataspace. The tape parameters are too flexible and not well documented. You can specify block size and number of 1k blocks. Without specific instructions, a novice SA can specify a writable parameter that is difficult to read. Are you getting the drift? And yes, we have lost a three year old, 200MB database because INFORMIX turbo cannot read the tapes that it made, and there is no other way to recover the data. dd will read the data and even place it on the hard disk, but without a complete understanding of the turbo engine, and how it manages the data, recovery of the data is VERY doubtfull. I have two copies of the database, both are five tapes long. Turbo will read the most current set, but the corruption that caused the crash, also exists on these tapes, so that restoring these tapes has the effect of restoring the corruption. The corruption is in the area where all of the systables are. Very similar to what happens when you lose an inode containing a directory entry. Using the public domain program bpatch, I can 'see' that my data is there. So close... The older set of tapes, will not be read by turbo. Nothing I do seems to help. And last of all, the kindly folks at INFORMIX are simpathetic, but will do nothing to help me. And how is your weekend going? B-) +=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=+ | Honesty is in the eye of | Phil Meyer | | the creditor. | Usenet: <backbone>!attctc!lodestar!phil | | | VoiceNet: (214) 991-0897 | +=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=+