On Thu, 16 May 2002, Taylor, ForrestX wrote: >Date: Thu, 16 May 2002 09:58:32 -0700 >From: "Taylor, ForrestX" <[EMAIL PROTECTED]> >To: [EMAIL PROTECTED] >Reply-To: [EMAIL PROTECTED] >Content-Type: text/plain; charset=us-ascii; format=flowed >Subject: Re: Testing Ext3 filesystem. > >Mark Cooke wrote: > >> Hi, >> I've just converted my whole system on my test server to Ext3. >> and I really after some adivse of testing this. >> >> Am I correct I the way I think Ext3 should work? >> >> If I'm moving say a 650 iso from /tmp /pub on an Ext3 system and then >> halfway thru I pull the plug. >> What should the result be after it boots backup? >> >> 1. Would the .journal note that it was trying to move a file and finish >> off the move. >> 2. Delete to destination file and leave the original in place. >> 3. I'm completely wrong and some one is going to correct me :-) >> >> >> Thanks in advance >> >> Mark > >Ext3 will not complete the activity that was happening when the power >was interrupted. It tries to bring the system back to a last known >state, but you may have corrupted files after (such as a file that you >were transferring). It does help in that it does not have to go through >the fsck check after an unclean shutdown (have you ever had to fsck an >80 GB disk??). > >Moving a file does not remove the source file until it has completed, >thus in your example, the destination file would not be finished, but >the source would be fully intact. You could probably use rsync to >continue moving the file, so that you wouldn't have to start over, but >I'm not sure how much of an issue that is. > >This is in the Reference guide for 7.3: >http://www.redhat.com/docs/manuals/linux/RHL-7.3-Manual/ref-guide/ch-ext3.html > >Michael K. Johnson wrote this: > >http://www.redhat.com/support/wpapers/redhat/ext3/
Another thing to note, is that journalled filesystems are NOT a substitute for backups. Journalled filesystems do NOT prevent filesystem corruption, nor data loss. Journalled filesystems DO however raise the bar on data integrity, and MINIMIZE the chance of data loss/corruption. Doing stupid things can still result in data loss on journalled filesystems. The filesystem ensures that the filesystem itself is in tact. ext3 has other modes of operation which trade off speed for further increased reliability, however the example given above will result in a half written file on any filesystem. That is an expected behaviour, and not a fault. In order to protect against such a failure simulated or otherwise, one would need to employ redundancy and HA, and use UPS backup power. Even that isn't 100% protection against problems. It's always a game of percentage juggling. If Bill Gates had a dime for every time Windows crashed, he'd be rich! _______________________________________________ Seawolf-list mailing list [EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/seawolf-list
