RE: Assert Failed: Buffer modified in inconsistent chunk

Dear Art, Thanks for replying. Allow me to seek a few clarifications from you. See below: -----Original Message----- From: Art S. Kagel [SMTP:kagel@bloomberg.net] Sent: Tuesday, 15 September, 1998 10:48 PM To: Richard W Sinyem Subject: Re: Assert Failed: Buffer modified in inconsistent chunk Richard W Sinyem wrote: > > Dear People, > > We are running I-Online 7.10.UD1 on HP-UX 10.01 and recently encountered the following error/problems: > [SNIP] > Could someone kindly explain the possible causes of the chunk to be inconsistent as we have > perform the oncheck a few days earlier before this incident. Furthermore, after restoration (and loss > of some data) the database is back to normal order. Additional preventive steps are also welcomed. One possibility is that someone added a new chunk that was a link to a device/file that was already being used by another chunk. Ospaces and onmonitor only check the filename (ie linkname) and do not follow the link to the FS/inode or Major/Minor device number to avoid overlap and --- indeed the second chunk WAS originally used as another dbspace. But that was a couple of months ago. What happens was that we, drop that dbspace and remove the link file. Then we re-use that chunk for the current dbspace. During, the Add-Chunk process, no error was reported and furthermore, during the last couple of months, the database engine is error free, (except for the last incident). clashes. This can also happen if you have multiple instances on the same machine since IDS has no way of checking between different instances. Since you have already restored the evidence is probably gone that this was the cause of the damage unless the links still exist. --- the link has been removed long time ago. In this case you still have a problem and the corruption will happen again, indeed it actually already has, by restoring the original chunk's data you have damaged the new chunk. --- what do you means by damaging the new chunk after the restoration ? If you have this problem get my utils_ak submission to IIUG software repository which has an onspaces wrapper script to prevent this. -- would this means that we need to rebuild the dbspace. Thanks and best regards. Art S. Kagel