Re: Could not bfget in Informix IDS 7.31

Hi, it is a bit difficult to tell in hindsight what has happened first, the chunk coming down or the corruption of data. You could maybe try to figure it out by comparing times in the IDS message log, possibly with times in AF-files ... I would assume that the chunks came down due to some serious problem with the I/O (disk crash ?, controller problem ?, ...). If the problem was transient (e.g. disk array without power during IDS startup), it would also have caused the chunks to come down. To fix it, you may have to repair some hardware. Afterwards, you probably should restore the dbspaces containing the down chunks. This can be a warm restore. For this you need a backup/archive from before the problem first appeared. With that most probably the corrupt data problem will also be fixed (unless a corruption existed already before the backup and was not detected by the backup, which woudl be a rather rare scenario). This is the normal way to fix such problems. If you do not (or cannot) go this way, then you probably should consider contacting IBM Informix Technical Support to find other possibilities to fix this. Regards, Martin -- Martin Fuerderer IBM Informix Development Munich, Germany Information Management owner-informix-list@iiug.org wrote on 05.10.2005 11:54:52: > Hi. > > I have a problem with some tables in IDS 7.31 whose data seem to be > corrupted. When I do a Check on the table mark_translation it appears: > > TBLspace data check for copyprod:euroadm.mark_translation > Table fragment in DBspace pem_data3 > Table fragment in DBspace pem_data2 > Table fragment in DBspace pem_data1 > Table fragment in DBspace pem_data4 > Could not bfget pagenum 3c50, iserrno 0 > > It involves that I cannot do a nested query: > > select * from mark_translation where ... in (..)> > As well I have some chunks DOWN so I have the following questions: > > 1) Is it only a register which is DOWN or it an entire PAGE? Is it > possible to know which are? By 3c50 I understand that the page/register > 15440 (3c50 in Decimal) is wrong ? > > 2) Is it possible to drop this registers corrupted? How to do it? > > 3) How has become this registers corrupted? Is it possible to control > that it does not happen any more? > > 4) The corrupted chunk provokes that the chunk where it is placed > becomes DOWN? or since the chunk is DOWN then the page where it is > stored becomes down too? > > 5) Which alternatives I have to solve the problem? > > Thank you all. > > Alan sending to informix-list