On Saturday 15 September 2007 10:16:49 pm Aaron Kulkis wrote: ... > > What year that happened Aaron? > > 2003-2005 timeframe.
There are many changes in a package changelog since than: 11 in 2006-2007, version 3.6.19-87 and 17 in 2003-2005. versions 3.6.4 to 3.6.19 not all are bug fixes. According to the ChangeLog in the reiserfs source, it seems that only SUSE guys are maintaining reiserfs since version 3.6.19 was out in 2004. ... > > As you can read in the article reiserfs is easy to signal panic if > > hardware is bad. Other file systems are not that easy at halting > > the system. Taking that running fsck.reiserfs under circumstances > > is equivalent to mkfs.reiserfs as proponents of ext3 like to underscore, > > signaling panic to often is not the best strategy, although I can't > > say that since I installed it again I had sudden kernel panic > > (computer freeze) ever. > > Same disk drives (IBM Deskstore SCSI-2) with ext3 / XFS had no problems. > Again, reading article, linked in previous posts, can help understand why one drive has more problems than the other. The hard disks have complex hardware, driven with complex firmware. Error in firmware makes drive to behave different than specified, which will produce errors that file system has to notice and do something about. File system drivers are not perfect. Sometimes they panic to early, don't try to go around the error, other time they ignore error and continue running. Just from fact that ext2 and xfs have no trouble with certain hard disk drive model doesn't make them right and reiserfs wrong. It can be both ways. The reiserfs is noticing error, but handling is just basic 'stop any operation' instead to mark sectors as bad and find some other place to write data, and ext3/xfs just ignores errors. I would have to reread the article to be able to answer questions: - who does ignore errors more often? - who has better error handling? but the thesis is 153 pages with 131 references and not really easy to read. In any case author Vijayan Prabhakaran, by now probably Doctor of Philosophy in Computer Sciences, made descent effort to analyze files systems and help readers of thesis to get better picture of problems and problem handling. -- Regards, Rajko. -- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
