Thanks Arne, with "recovery" I meant log and data of course so that you shouldn't have any data loss. Nevertheless with "select * from roots where root = 50956" you will find the index and the table. The -9000 error does not cause such problems. This is a different problem. Normally page errors come from hardware errors (disc or controller) but I will have a look into the dumped page if it could be a software bug.
kind regards, Uwe >-----Original Message----- >From: Arne Gehlhaar [mailto:[EMAIL PROTECTED] >Sent: Monday, August 23, 2004 3:16 PM >To: Hahn, Uwe >Cc: [EMAIL PROTECTED] >Subject: Re: DB crash > > >Hi Uwe, > >here are the files as requested. The OS we are using is Red Hat Linux >release 7.3 (Valhalla). > >A database verify is done every night before a complete data >backup. No >error is reported in the logs. However after the complete data >backup we >also do an update of the statistics. This produces the folowing error: > >ERR >-24988,ERR_SQL: sql error >-9000,System error: Not yet implemented > >Does that have anything to do with the crash? > >The error first occurred on August 6th, so a recovery of an older dump >is not an option (more than 2 weeks data-loss). > >Is it possiple to find out which index or which table is >involved, so we >could drop and recreate those index(es)? > >Thanks for the help, >Arne > >Hahn, Uwe wrote: > >> Hi Arne, >> >> illegal entrypos means that you have a corrupt data page. >> In this case it is a page which belongs to an index. >> If the page iscorrupted on the data volume the db will crash >> everytime that page should be changed. >> >> In your rundirectory should exist a file d180201.cor. >> Please send it to me with your knldiag.err. (which os are you using?) >> >> You should check the database (verify/check data). >> If the page error is detected by this check you >> can try to drop and create the index again. >> >> Another choice is to to take an older data backup which >> must be checked too if there are all pages correct and >> to recover. >> >> kind regards >> Uwe >> >> >> >>>-----Original Message----- >>>From: Arne Gehlhaar [mailto:[EMAIL PROTECTED] >>>Sent: Monday, August 23, 2004 1:16 PM >>>To: [EMAIL PROTECTED] >>>Subject: DB crash >>> >>> >>>Hi, >>> >>>in the past couple of days we have experienced a number of database >>>crashes which we at first believed to be connected to incremental >>>database dumps (PAGES). However tests have shown that this >is not the >>>case. We now suspect normal UPDATE/DELETE operations to be the cause. >>> >>>We are using Kernel Version 7.4.3 Build 010-00-035-462 >>> >>>Below is a extract of the knldiag.err >>> >>>Any hints will be appreciated. >>> >>>Thanks in advance, >>>Arne >>> >>> >>>-- >>>Institut fuer Seeverkehrswirtschaft und Logistik http://www.isl.org/ >>Arne Gehlhaar mailto:[EMAIL PROTECTED] >>Universitaetsallee GW1 Block A tel:+49 421 22096-49 >>D-28359 Bremen fax:+49 421 22096-55 >> >> >>-- >>MaxDB Discussion Mailing List >>For list archives: http://lists.mysql.com/maxdb >>To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED] >> > > -- Institut fuer Seeverkehrswirtschaft und Logistik http://www.isl.org/ Arne Gehlhaar mailto:[EMAIL PROTECTED] Universitaetsallee GW1 Block A tel:+49 421 22096-49 D-28359 Bremen fax:+49 421 22096-55 -- MaxDB Discussion Mailing List For list archives: http://lists.mysql.com/maxdb To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
