Editing RAID Cards (section)

= 3ware =

== Notes ==

To replace a failed mirror with the drive that failed, boot up with only the failed drive (to be safe, make sure the other drive boots fine), delete the mirror on the bad drive. Boot up again with both drives and select the failed (now empty/available drive) and the remaining degraded mirror and select to rebuild. Hit F8 on the way out to save/rebuild.

*adding mirror(s) to a new/empty machine (with no os)
*adding mirror(s) to a running machine
*moving mirror(s) to a new/empty machine (with no os)
*moving mirror(s) to a running machine
*replacing dead/degraded mirror on running machine (please include pointers on 
which spare drives can be used as replacements*)
*silencing alarms (a nod to castle staff)

*please include documentation for both CLI and BIOS apps. where appropriate and 
where differences exist.
*include instructions up to the point where the partition is mounted and usable 
(or recognized by os in the case of new install)
*please include (separate) instructions for clearing off used drive so it can 
be reused in above procedures.

Tech support: 800 840-6055


== CLI (9xxx) ==
=== Replacing a failed drive ===


<pre>
tw_cli /c0 show all

Port   Status           Unit   Size        Blocks        Serial
---------------------------------------------------------------
p0     OK               u0     1.82 TB     3907029168    WD-WCAVY0647904
p1     OK               u0     1.82 TB     3907029168    WD-WCAVY0608298
p2     OK               u1     1.82 TB     3907029168    WD-WCAVY0629856
p3     OK               u1     1.82 TB     3907029168    WD-WCAVY0627316
p4     OK               -      1.82 TB     3907029168    WD-WCAVY0564054
</pre>

then you should run:

<pre>
tw_cli /c0 rescan
tw_cli /c0 show all

Port   Status           Unit   Size        Blocks        Serial
---------------------------------------------------------------
p0     OK               u0     1.82 TB     3907029168    WD-WCAVY0647904
p1     OK               u0     1.82 TB     3907029168    WD-WCAVY0608298
p2     OK               u1     1.82 TB     3907029168    WD-WCAVY0629856
p3     OK               u1     1.82 TB     3907029168    WD-WCAVY0627316
p4     DEGRADED         u1     1.82 TB     3907029168    WD-WCAVY0564054
</pre>

and the array should show that it is rebuilding

BUT, if you attempt to rebuild with a command like that, and you get:

Error: The following drive(s) cannot be used [4].

you need to remove it, then rescan, then rebuild:

 tw_cli /c0/p4 remove
 tw_cli /c0 rescan
 tw_cli /c0/u0 start rebuild disk=4

== Rebuild speed ==
Do the following to slow down a rebuild speed so as to improve performance

(1=fast, 5=slow)
 tw_cli /c0 set rebuild=5

turn off storsave protection (set back to protect)
 tw_cli /c0/u0 set storsave=perform

turn off write cache
 tw_cli /c0/u0 set cache=off

== Hot Spare ==

To add a hot spare drive
  tw_cli /c0 add type=spare disk=XX

== CLI (8xxx) ==

Reference guide: ftp://ftp.rackable.com/public/Technical%20Support/Pdf%20files/3Ware/7000_8000/CLI-UserGuide.pdf

=== Replacing a failed drive ===

If you don't remove the drive from a RAID-1 array before you replace it, the new drive comes up
as a separate JBOD device.  To fix this do

<pre>
tw_cli maint deleteunit c0 u1
tw_cli maint remove c0 u0 p1          (may not be needed)
tw_cli maint rescan c0
tw_cli /c0 show all
tw_cli maint rebuild c0 u0 p1
tw_cli /c0 show all
</pre>

After replacing the 2 dead drives, we see:

<pre>
tw_cli info c0
Controller: c0
-------------
Driver:   1.50.01.002
Model:    7500-8
FW:       FE7X 1.05.00.068
BIOS:     BE7X 1.08.00.048
Monitor:  ME7X 1.01.00.040
Serial #: F11605A3180172
PCB:      Rev3
PCHIP:    1.30-33
ACHIP:    3.20


# of units: 3
        Unit 0: JBOD 186.31 GB ( 390721968 blocks): OK
        Unit 1: RAID 5 465.77 GB ( 976790016 blocks): DEGRADED
        Unit 5: RAID 5 698.65 GB ( 1465185024 blocks): DEGRADED

# of ports: 8
        Port 0: WDC WD2000JB-00KFA0 WD-WCAMT1451690 186.31 GB (390721968 blocks): OK(unit 0)
        Port 1: WDC WD2500JB-00GVC0 WD-WCAL78219488 232.88 GB (488397168 blocks): OK(unit 1)
        Port 2: WDC WD2000  0.00 MB (0 blocks): OK(NO UNIT)
        Port 3: WDC WD2500JB-00GVC0 WD-WMAL73882417 232.88 GB (488397168 blocks): OK(unit 1)
        Port 4: WDC WD2000  0.00 MB (0 blocks): OK(NO UNIT)
        Port 5: WDC WD2500JB-00GVA0 WD-WMAL71338097 232.88 GB (488397168 blocks): OK(unit 5)
        Port 6: WDC WD2500JB-32EVA0 WD-WMAEH1301595 232.88 GB (488397168 blocks): OK(unit 5)
        Port 7: WDC WD2500JB-00GVC0 WD-WCAL78165566 232.88 GB (488397168 blocks): OK(unit 5)
</pre>

which is to say there are 2 drives which don't belong to any units and don't have any size. They need to be removed from the controller:
 tw_cli maint remove c0 p2
 tw_cli maint remove c0 p4

although it seems like we should be able to add them back in with
 tw_cli maint add c0 p2 jbod
 tw_cli maint add c0 p2 spare
it doesn't work, and there's no rescan command, so we resort to rebooting.

We won't be able to boot all the way up however, cause the 2 drives will turn into JBOD's and throw off the device ordering:
<pre>twed0: <Unit 0, JBOD, Normal> on twe0
twed0: 190782MB (390721968 sectors)
twed1: <Unit 1, RAID5, Degraded> on twe0
twed1: 476948MB (976790016 sectors)
twed2: <Unit 2, JBOD, Normal> on twe0
twed2: 239372MB (490234752 sectors)
twed3: <Unit 4, JBOD, Normal> on twe0
twed3: 239372MB (490234752 sectors)
twed4: <Unit 5, RAID5, Degraded> on twe0
twed4: 715422MB (1465185024 sectors)
twed5: <Unit 0, RAID5, Normal> on twe1
twed5: 715422MB (1465185024 sectors)
twed6: <Unit 4, RAID5, Normal> on twe1
twed6: 715422MB (1465185024 sectors)
</pre>

So whereas twed2 used to be a RAID5 device, it got pushed down (to twed4?) by the JBOD standalone drive. So we could edit the fstab and stop the OS from trying to mount the other devices. After the rebuild (or during) we could reboot and the next time it comes up the devices will fall back to their regular ordering. 
We usually opt to, over serial console, enter into single user mode (which is automatic with the failed mounts), and do the remirroring:

<pre>
mount /dev/twed0s1g           /usr
(tw_cli is in /usr we we need to mount it manually)

tw_cli info c0
Controller: c0
-------------
Driver:   1.50.01.002
Model:    7500-8
FW:       FE7X 1.05.00.068
BIOS:     BE7X 1.08.00.048
Monitor:  ME7X 1.01.00.040
Serial #: F11605A3180172
PCB:      Rev3
PCHIP:    1.30-33
ACHIP:    3.20


# of units: 5
        Unit 0: JBOD 186.31 GB ( 390721968 blocks): OK
        Unit 1: RAID 5 465.77 GB ( 976790016 blocks): DEGRADED
        Unit 2: JBOD 233.76 GB ( 490234752 blocks): OK
        Unit 4: JBOD 233.76 GB ( 490234752 blocks): OK
        Unit 5: RAID 5 698.65 GB ( 1465185024 blocks): DEGRADED

# of ports: 8
        Port 0: WDC WD2000JB-00KFA0 WD-WCAMT1451690 186.31 GB (390721968 blocks): OK(unit 0)
        Port 1: WDC WD2500JB-00GVC0 WD-WCAL78219488 232.88 GB (488397168 blocks): OK(unit 1)
        Port 2: WDC WD2500SB-01RFA0 WD-WMANK3040813 233.76 GB (490234752 blocks): OK(unit 2)
        Port 3: WDC WD2500JB-00GVC0 WD-WMAL73882417 232.88 GB (488397168 blocks): OK(unit 1)
        Port 4: WDC WD2500SB-01RFA0 WD-WMANK3356318 233.76 GB (490234752 blocks): OK(unit 4)
        Port 5: WDC WD2500JB-00GVA0 WD-WMAL71338097 232.88 GB (488397168 blocks): OK(unit 5)
        Port 6: WDC WD2500JB-32EVA0 WD-WMAEH1301595 232.88 GB (488397168 blocks): OK(unit 5)
        Port 7: WDC WD2500JB-00GVC0 WD-WCAL78165566 232.88 GB (488397168 blocks): OK(unit 5)
</pre>

2 new drives show up as JBOD's, which is fine, we can allocate into a mirror, but first we must delete the JBOD unit:
BE VERY CAREFUL about this, double check that the unit you are deleting has 1 member and it's member is on the port which contains the new drive.
<pre>
tw_cli maint deleteunit c0 u2
Deleting unit /ct0/u2 ...wed2: detached
Done.

tw_cli maint deleteunit c0 u4
Deleting unit /ct0/u4 ...wed3: detached
Done.

tw_cli info c0
Controller: c0
-------------
Driver:   1.50.01.002
Model:    7500-8
FW:       FE7X 1.05.00.068
BIOS:     BE7X 1.08.00.048
Monitor:  ME7X 1.01.00.040
Serial #: F11605A3180172
PCB:      Rev3
PCHIP:    1.30-33
ACHIP:    3.20


# of units: 3
        Unit 0: JBOD 186.31 GB ( 390721968 blocks): OK
        Unit 1: RAID 5 465.77 GB ( 976790016 blocks): DEGRADED
        Unit 5: RAID 5 698.65 GB ( 1465185024 blocks): DEGRADED

# of ports: 8
        Port 0: WDC WD2000JB-00KFA0 WD-WCAMT1451690 186.31 GB (390721968 blocks): OK(unit 0)
        Port 1: WDC WD2500JB-00GVC0 WD-WCAL78219488 232.88 GB (488397168 blocks): OK(unit 1)
        Port 2: WDC WD2500SB-01RFA0 WD-WMANK3040813 233.76 GB (490234752 blocks): OK(NO UNIT)
        Port 3: WDC WD2500JB-00GVC0 WD-WMAL73882417 232.88 GB (488397168 blocks): OK(unit 1)
        Port 4: WDC WD2500SB-01RFA0 WD-WMANK3356318 233.76 GB (490234752 blocks): OK(NO UNIT)
        Port 5: WDC WD2500JB-00GVA0 WD-WMAL71338097 232.88 GB (488397168 blocks): OK(unit 5)
        Port 6: WDC WD2500JB-32EVA0 WD-WMAEH1301595 232.88 GB (488397168 blocks): OK(unit 5)
        Port 7: WDC WD2500JB-00GVC0 WD-WCAL78165566 232.88 GB (488397168 blocks): OK(unit 5)
</pre>

Now the 2 drives show up as full size and available. We can rebuild:

<pre>
tw_cli maint rebuild c0 u1 p2
Rebuild started twon unit /c0/u1
AEN: <twed1: rebuild started>

tw_cli maint rebuild c0 u5 p4
Rebuild started ton unit /c0/u5
AEN: <twed1: rebuild started>

tw_cli info c0
Controller: c0
-------------
Driver:   1.50.01.002
Model:    7500-8
FW:       FE7X 1.05.00.068
BIOS:     BE7X 1.08.00.048
Monitor:  ME7X 1.01.00.040
Serial #: F11605A3180172
PCB:      Rev3
PCHIP:    1.30-33
ACHIP:    3.20


# of units: 3
        Unit 0: JBOD 186.31 GB ( 390721968 blocks): OK
        Unit 1: RAID 5 465.77 GB ( 976790016 blocks): REBUILDING (0%)
        Unit 5: RAID 5 698.65 GB ( 1465185024 blocks): REBUILDING (0%)

# of ports: 8
        Port 0: WDC WD2000JB-00KFA0 WD-WCAMT1451690 186.31 GB (390721968 blocks): OK(unit 0)
        Port 1: WDC WD2500JB-00GVC0 WD-WCAL78219488 232.88 GB (488397168 blocks): OK(unit 1)
        Port 2: WDC WD2500SB-01RFA0 WD-WMANK3040813 233.76 GB (490234752 blocks): OK(unit 1)
        Port 3: WDC WD2500JB-00GVC0 WD-WMAL73882417 232.88 GB (488397168 blocks): OK(unit 1)
        Port 4: WDC WD2500SB-01RFA0 WD-WMANK3356318 233.76 GB (490234752 blocks): OK(unit 5)
        Port 5: WDC WD2500JB-00GVA0 WD-WMAL71338097 232.88 GB (488397168 blocks): OK(unit 5)
        Port 6: WDC WD2500JB-32EVA0 WD-WMAEH1301595 232.88 GB (488397168 blocks): OK(unit 5)
        Port 7: WDC WD2500JB-00GVC0 WD-WCAL78165566 232.88 GB (488397168 blocks): OK(unit 5)
</pre>

And now we can reboot, the rebuild will continue backgrounded, and the devices will come up in correct order:

<pre>twed0: <Unit 0, JBOD, Normal> on twe0
twed0: 190782MB (390721968 sectors)
twed1: <Unit 1, RAID5, Rebuilding> on twe0
twed1: 476948MB (976790016 sectors)
twed2: <Unit 4, RAID5, Rebuilding> on twe0
twed2: 715422MB (1465185024 sectors)
twed3: <Unit 0, RAID5, Normal> on twe1
twed3: 715422MB (1465185024 sectors)
twed4: <Unit 4, RAID5, Normal> on twe1
twed4: 715422MB (1465185024 sectors)</pre>