-----Forwarded Message----- >From: Jeff Johnson <n3...@mac.com> >Sent: Feb 22, 2011 8:14 PM >To: coo...@mandrivalinux.org >Subject: [Cooker] Upgrding db-5.1.19 -> db-5.1.26 needs full rebuilds > >(cool! a bug related to RPM might have been found _BEFORE_ it ended up in >cooker-devel e-mail ;-) > >I got a bug report from bero at Arklinux with these symptoms: > >Preparing... ########################################### [100%] >rpmdb: not a restored transaction >rpmdb: PANIC: Invalid argument >==> rpmdbe_event_notify(0x9833aa8, PANIC(0), 0xbfe2b5d4) app_private (nil) >rpmdb: PANIC: fatal region error detected; run recovery >==> rpmdbe_event_notify(0x9833aa8, PANIC(0), 0xbfe2b6cc) app_private (nil) > 1:xorg ########################################### [ 20%] >rpmdb: PANIC: fatal region error detected; run recovery >==> rpmdbe_event_notify(0x9833aa8, PANIC(0), 0xbfe2b5bc) app_private (nil) > >I've never ever seen that 1st Berkeley DB error before. > >A quick check with Google found this report: > http://forums.oracle.com/forums/thread.jspa?threadID=2175221&tstart=0 >and I just confirmed with jcea that upgrading to db-5.1.26 without rebuilding >everything that uses/needs > #include <db.h> >will lead to "not a restored transaction". > >Presumably some #define got changed, breaking (for the 1st time ever afaik) >the typical expectation that 5.1.19 -> 5.1.26 is just bug fix patches. > >Disclaimer: > This is still just 2nd hand info because I haven't personally looked. > >So -- if you go to db-5.1.26 -- plan on rebuilding RPM and whatever >else is linking -ldb-5.1. > >There's nothing critically important in db-5.1.26 (or I would remember). > >It would be nice to get on the other side of this minor "speed bump" >in Berkeley DB by 2011.0 release. But db-5.1.19 is known "gud enuf" by now. > >hth > >73 de Jeff >
smime.p7s
Description: S/MIME cryptographic signature