[linux-lvm] online resize

[Zac Slade]

On Sunday 26 March 2006 06:35, Jan-Benedict Glaw wrote:
> Nice story, but as a bug report, this is mostly useless:
This is not a bug report.  I did not file one (at the time I didn't have the 
time to track it).  And I needed my data then.  The time it would take to 
track the bug down, get a fix and get it running would have put me even 
further behind.  I remade the fs into reiser3 and restored the data.  I have 
had many successful resizes (but grow and shrink) since using reiser3.

>   * After umount, did you run an e2fsck -f?
Yes.

>   * Do you have a script(1) log of the whole session? (You'd probably
>     have this started as a last resort log then you started the
>     fsck'ing after the resize.)
No.  The session was not run through script(1).

>   * What exactly is "ext resizing utilities"? Did you save a copy of
>     it (incl. sources to find out about additionally applied patches?)
resize2fs.  I do not have a copy of the version used.

>   * These nonsensical names you refer to are the file's inode numbers,
>     in decimal. For mission-critical systems, it's wise to regularly
>     get inode listings with filenames. Did you compare those inode
>     numbers with your backups?
I know they are inode numbers, but they don't help you much in that situation.  
You think people keep inode listings with backups?  What a perfect world you 
live in.  I think it's not superflous information, but it is not a part of 
your average backup.  As a matter of fact I've not incountered a system that 
does this.  Sure you can do this on your own and it sort of makes sense.  
This just seems like a strange suggestion to me.

>   * Did you prepare an image-backup of your 1GB container beforehand?
>     Helps for easy recovery as well as error reproduction (esp.
>     because 1GB isn't all that hard to store on one CD when there's
>     plenty of free space, which can be made to compress very well.)
I did restore the data eventually.  I apologize if that was not clear.  I was 
frustrated that the tool did not perform as expected and it is something I 
should have followed up on with a bug report.  I should go back and do some 
experimenting with the current e2fsprogs and see if they behave better now.

-- 
Zac Slade
krakrjak at volumehost.net
ICQ:1415282 YM:krakrjak AIM:ttyp99