Re: Failed consistency check

Had same problem ! ian.goddard@geo2.poptel.org.uk (Ian Goddard) wrote: >In article <449iip$cqv@newssvr.med.osd.mil> zsun@a000dbss.med.osd.mil(Zhenling Sun) writes: >>Subject: Failed consistency check >>From: zsun@a000dbss.med.osd.mil(Zhenling Sun) >>Date: 26 Sep 1995 18:59:37 GMT > >>When I do tbcheck -pt on one of our databases, I got following error in >>online.log file: >> newmode: partnum 0x3000028, page 0x6002074, invalid type 0x9 >> -- Fail Consistency Check -- newmode -- pid 3213 user=200 us=80401bb0 > >>user 200 is informix id, pid 3213 is the process tbcheck -pt, tbcheck revealed >>that table 0x3000028 has 8 pages allocated, and 3 pages are used, physical page >>address for this table starts from 0x3014e1. A tbcheck -pP showes that the >>trouble starts at page 0x3014e4 ( the first free page of this table), the >>address in the page header is "0x6002074", instead of "0x3014e4". > >>I tried to unload this table, drop this table, recreate and re-load this table, >>and tbcheck -pt will generate same error message. If I do a dbexport the whole >>database, drop the database, then dbimport the database, this error will just >>go to another table. > >>Has anyone see this error before? I am thinking of dbexport the database, then >>drop the chunk, reinitialize the chunk, maybe then the error will go away? >>Maybe someone out there has a better solution? I will appreciate any help. >>email or post are fine. > > >We get occasional messages like that in 4.11 (I presume your engine is DSA). >Tech support suggested it may be corruption in shared memory. We have had no >evidence of harm done and tend to ignore them now but they seem to be >transient, unlike your example. Following the possibility that it is in >shared memory, however, have you tried stopping and restarting OnLine? > >Ian We had a chunk defined with 16 Gb offset on a 2 Gb rlv. This resulted in a wrap around, leaving the chunk freelist page in the middle of another chunk. A table on this other chunk showed your Error. You might check this with onstat -d Good luck !