vinum raid5 subdisks keep changing length?

[Tony Frank]

Hi,

On Tue, Feb 17, 2004 at 09:51:30AM +1030, Greg 'groggy' Lehey wrote:
> On Monday, 16 February 2004 at 22:04:44 +1100, Tony Frank wrote:
> > Running vinum on RELENG_4.
> >
> > No changes to source, vinum loaded as a KLD as per instructions in handbook &
> > man pages.
> >
> > As per the subject I am having what appears to be strange behaviour with vinum.
> >
> > After a number of teething problems I eventually setup a vinum root for
> > my system which covers / swap /var and /usr.
> > This works just fine, though I am yet to experiment with failure scenarios.
> >
> > However I also configured a raid5 volume and may be having issues with it.
> > Currently just a test system and no obvious data loss but I have some
> > concerns before I put it into production.
> 
> > Specifically I originally configured five subdisks (on different
> > drives) with identical lengths and then combined them into a raid5
> > plex.  This worked ok and I successfully initialised the plex and
> > used newfs on it.  Successfully able to read&write to the volume
> > (1G/14G used at the moment)
> >
> > The original config looked as so:
> >
> > volume data
> > plex name data.p0 org raid5 984s vol data
> > sd name data.p0.s0 drive eightgig plex data.p0 len 8802864s driveoffset 6335201s plexoffset 0s
> > sd name data.p0.s1 drive vinumdrive0 plex data.p0 len 8802864s driveoffset 265s plexoffset 984s
> > sd name data.p0.s2 drive vinumdrive1 plex data.p0 len 8802864s driveoffset 265s plexoffset 1968s
> > sd name data.p0.s3 drive vinumdrive2 plex data.p0 len 8802864s driveoffset 265s plexoffset 2952s
> > sd name data.p0.s4 drive vinumdrive3 plex data.p0 len 8802864s driveoffset 265s plexoffset 3936s
> >
> > Now after each system reboot I see messages on system console from vinum.
> > (they are not logged in vinum_history or messages)
> 
> These messages should appear in /var/log/messages.  Have you changed
> your syslogd.conf?

No, default 4.9 config file, the vinum messages do not show up in 'dmesg' either.
I have changed to include the 'all.log' and 'console.log' in the syslogd.conf so
I will see if that changes this behaviour.

> > vinum: updating configuration from /dev/ad0s1h
> > #(repeated for all 6 drives)
> > vinum: removing 2560 blocks of partial stripe at the end of data.p0
> 
> Yes, this is a feature, not a bug.

While I accept this for a striped plex made up with subdisk of unequal length, 
I had already taken that into account, hence the original subdisk length:
 8946 * 984 = 8802864

As I understand it, this should have resulted in a whole number of stripes.
I have 5 subdisks so I'm not sure if I should have factored that into my calculations?

> > A "vinum printconfig" now shows difference values for the subdisks.
> > This value has changed after each reboot.
> > Current relevant parts of 'printconfig' are:
> >
> > volume data
> > plex name data.p0 org raid5 984s vol data
> > sd name data.p0.s0 drive eightgig plex data.p0 len 8799978s driveoffset 6335201s plexoffset 0s
> > sd name data.p0.s1 drive vinumdrive0 plex data.p0 len 8799486s driveoffset 265s plexoffset 984s
> > sd name data.p0.s2 drive vinumdrive1 plex data.p0 len 8799158s driveoffset 265s plexoffset 1968s
> > sd name data.p0.s3 drive vinumdrive2 plex data.p0 len 8798420s driveoffset 265s plexoffset 2952s
> > sd name data.p0.s4 drive vinumdrive3 plex data.p0 len 8797600s driveoffset 265s plexoffset 3936s
> 
> This is a bug, not a feature.
> I'll try to take a look at it today.  

Please advise if I can help - I have plenty of free time this week and am willing to get my hands dirty.

Another item (which I think I read about somewhere but can no longer find the reference)
is that after a reboot the vinum drives all reference da0h ad2h etc rather than the
original value of ad2s1h da0s1h.  Also after the reboot, each drive is shown as 100% free which is a concern.

Thanks,

Tony