amd64/103043: graid3 does not rebuild degraded arrays

[douglas steinawnd]

>Number:         103043
>Category:       amd64
>Synopsis:       graid3 does not rebuild degraded arrays
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-amd64
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Fri Sep 08 16:20:14 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     douglas steinawnd
>Release:        6.1-STABLE
>Organization:
>Environment:
FreeBSD paris.fx.org 6.1-STABLE FreeBSD 6.1-STABLE #2: Thu Sep  7 06:52:51 PDT 2006     root at paris.fx.org:/usr/obj/usr/src/sys/HAWK6-SMP  amd64
>Description:
geom_raid3 is not rebuilding stale components. Removing/inserting, using "graid3 rebuild", etc. commands doesn't help. Syslog shows that rebuild has started, but the synchronized value stays at 0%:

GEOM_RAID3: Device 0: provider ad12 detected.
GEOM_RAID3: Device 0: rebuilding provider ad12.

dzs at paris:~$ graid3 list
Geom name: 0
State: DEGRADED
Components: 5
Flags: NOAUTOSYNC, VERIFY
GenID: 1
SyncID: 21
ID: 529755383
Zone64kFailed: 5947
Zone64kRequested: 241897
Zone16kFailed: 1671
Zone16kRequested: 76252
Zone4kFailed: 138407
Zone4kRequested: 5499275
Providers:
1. Name: raid3/0
   Mediasize: 1600353826816 (1.5T)
   Sectorsize: 2048
   Mode: r5w4e4
Consumers:
1. Name: ad4
   Mediasize: 400088457216 (373G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 21
   Number: 0
   Type: DATA
2. Name: ad6
   Mediasize: 400088457216 (373G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 21
   Number: 3
   Type: DATA
3. Name: ad8
   Mediasize: 400088457216 (373G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 21
   Number: 1
   Type: DATA
4. Name: ad10
   Mediasize: 400088457216 (373G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 21
   Number: 2
   Type: DATA
5. Name: ad12
   Mediasize: 400088457216 (373G)
   Sectorsize: 512
   Mode: r1w1e1
   State: SYNCHRONIZING
   Flags: DIRTY, SYNCHRONIZING, FORCE_SYNC
   GenID: 1
   SyncID: 21
   Synchronized: 0%
   Number: 4
   Type: PARITY



>How-To-Repeat:
Build a new graid3 array (perhaps using md devices), remove one component, then insert it again. Rebuild appears to start, but monitoring with "gstat" shows no real activity. Output of "graid3 list" or "graid3 status" shows 0%.
>Fix:
Revert to rev 1.40.2.12 of src/sys/geom/raid3/g_raid3.c. This problematic change was apparently MFC'd on Sept 4th.
>Release-Note:
>Audit-Trail:
>Unformatted: