Re: bad btree node

Naomi Walker writes: |> We have had some hardware problems, which may or may not be fixed (AST 33pc). |> One table in particular has problems when we try to insert a row. We get |> a -271 error (could not insert row into table). I could add a row thru isql, |> but attempts to retrieve it caused isql to core dump in a malloc call. |> |> I unloaded the table and rebuilt it, but I am still getting the -271 error, |> along with the following yukky stuff in the Online error log. How did you unload it? |> |> We have run a series of tbchecks, but they show no problems (cd, ci, ce, cr). |> [ consistency check errors in the log file ] |> |> My only choice now, I think, is to tbunload the databases in the partition, |> initialize, and put it all back. I'm worried the tbunload will pick up the |> bad goop (good tekkie talk, huh). I may try anyway, out of desperation. |> Tbunload, because it unloads the data and index pages, will retain the corruption. I wouldn't bother. |> What now? What should one do when you get bad btree nodes? I do not have |> enough unix file system space to unload and reload, and it would take hours |> and hours, close to days. This is a production system. Arggghhhh! I suggest dropping the indexes on that table and then rebuilding them. The message indicates that the page has a corrupted header. While it is impossible to know for sure whether the corruption is more widespread or not (a tbcheck -cD should be run), if it is limited to the index pages, then rebuilding the indexes should take care of things. If, however, your data pages are affected as well, you may be best off digging up your archives. If the corruption was the result of hardware problems, it could be very widespread. Dave Disclaimer: These opinions are not those of Informix Software, Inc. ************************************************************************** "I look back with some satisfaction on what an idiot I was when I was 25, but when I do that, I'm assuming I'm no longer an idiot." - Andy Rooney