RAID manager failed!!!!!!!!!!!

To save others the time to answer again, please review your old thread

[url=http://forum.sun.com/jive/thread.jspa?threadID=96533&messageID=331344# 331344]connect to StorEdge A1000 [/url]

Have you installed the required HVD SCSI hba (SYM22802) in the meantime ?

What version of Solaris ?

any advice?

Sure, re-read you old posts.

Michael

MAALATFT at 2007-7-7 0:27:10 >

Hello,

the A1000 isn't supported under Solaris 10 (by Sun, under service contract, ...), but works.

Rainer Orth and Fredrik Lundholm posted instructions on comp.unix.solaris. This an except from the post

****************************************************************

Fredrik Lundholm < ... > wrote:

> RDAC stands for Redundant Disk Array Controller Support, and is only

> needed for path failover on RSM2000/A3000/A3500/A3500FC. Since A1000

> only has a single controller it is not needed. rm6 talks to the A1000

> through Lun 0.

I had complained loudly during Solaris 10 Platinum Beta when I first

learned about removal of RM6/A1000 support in S10 FCS *without any

appropriate prior notice (i.e.: EOL notices) in S9 Release Notes*.

Unfortunately, this didn't help to bring the support back, but my support

engineer quickly determined that RM6 runs just fine with an A1000.

I later learned that the claimed reason for support removal was that the

A1000 itself had been EOLed before S10 went FCS.

> Perhaps in some cases RM6 is looking for the RDAC paths in order to

> determine where the modules are? So while the driver isn't technically

> required it can still cause issues by its absence?

The rdriver won't load in S10:

Nov 21 23:24:47 sequoia krtld: [ID 819705 kern.notice] /kernel/drv/sparcv9/rdriver: undefined symbol

Nov 21 23:24:47 sequoia krtld: [ID 826211 kern.notice] 'dev_get_dev_info'

Nov 21 23:24:47 sequoia krtld: [ID 472681 kern.notice] WARNING: mod_load: cannot load module 'rdriver'

Nov 21 23:24:47 sequoia genunix: [ID 370176 kern.warning] WARNING: forceload of drv/rdriver failed

Nov 21 23:24:47 sequoia krtld: [ID 819705 kern.notice] /kernel/drv/sparcv9/rdriver: undefined symbol

Nov 21 23:24:47 sequoia krtld: [ID 826211 kern.notice] 'dev_get_dev_info'

Nov 21 23:24:47 sequoia krtld: [ID 472681 kern.notice] WARNING: mod_load: cannot load module 'rdriver'

But since rdriver and rdnexus are only needed for RDAC (which doesn't do anything useful with the single-controller A1000), I just rem_drv'ed both and disabled RDAC support in /etc/osa/rmparams,

as I had already done in S8:

rmparams.rm6221Mon Nov 21 13:29:22 2005

+++ rmparamsTue Nov 22 00:23:58 2005

@@ -80,8 +80,9 @@

# RDAC driver/Resolution Daemon Failover parameters

-Rdac_SupportDisabled=FALSE

-Rdac_SinglePathSupportDisabled=FALSE

+# Disabled, useless on A1000 without redundant controllers

+Rdac_SupportDisabled=TRUE

+Rdac_SinglePathSupportDisabled=TRUE

Rdac_HotAddDisabled=PARTIAL

Rdac_HotAddIDs:4:5:

Rdac_NativeScsiDrivers:sd:

This way, an A1000 works just fine when connected to an E3000 with 4 x 167 MHz UltraSPARC I CPUs running Solaris Express Build 33, including arraymon and the RM6 GUI and CLI interfaces :-)

I'm actually using it for some ZFS testing, which works amazingly well, considering the old and slow hardware.

Hope this helps.

Rainer

-

Rainer Orth, Faculty of Technology, Bielefeld University

****************************************************************************

The instructions worked for me. I did remove the two forceloads (rdriver, rdnexus) in /etc/system as well.

Sideeffect: These changes removed the long delay (about 5-10 minutes) during boot Solaris 9/Solaris 10 too.

Michael

MAALATFT at 2007-7-7 0:27:10 >

Hello,

- check the cable and terminator (HVD), use the screws to attach them

tightly!

- issue probe-scsi all at the ok-prompt

Is the A1000 visible ?

If yes, reboot the system with boot -rv, capture the output.

If not, review the A1000 LEDs (--> A1000 User's Guide). Maybe you

have to replace the cable and/or terminator or hba.

Have these components pulled from a working system (are known

good) ?

- At the os level verify that all requires packages and patches (->

RM Installation Guide) are installed. Check the entries in /etc/system.

Michael

MAALATFT at 2007-7-7 0:27:10 >

Hello,

# iostat -En

it gave me this..

c4t0d0 Soft Errors: 1 Hard

Errors: 450 Transport Errors: 0

Vendor: Symbios Product: StorEDGE A1000 Revision:

0205 Serial No:

Size: 0.00GB <0 bytes>

Media Error: 0 Device Not Ready: 450 No Device: 0

Recoverable: 0

Illegal Request: 1 Predictive Failure Analysis: 0

then i did probe-scsi-all: it showed me nothing!!

probe-scsi-all is a command entered at the ok-prompt.

If iostat -En at the os level displays information for the A1000, the device is indeed detected, devices-nodes have been created, ...

Did you check the hardware, replaced components with spares ?

The reason that none of the installed disks are displayed might be that the A1000 isn't configured (no LUNs created ).

Vendor: Symbios Product: StorEDGE A1000 Revision: 0205.

The A1000 firmware is too old (2.05) !! RM 6.2 (and the drivers from this package) have firmware 3.01 as a prequisite.

The procedure to update the firmware and the required firmware levels/patches can be found on the following page (Sun StorEdge RAID Manager 6.22.1 Release Notes for Sun StorEdge A1000, A3000, A3500, and A3500FC Subsystems, Part-No. 805-7758-14).

http://www.sun.com/products-n-solutions/hardware/docs/Network_Storage_Solutions /Workgroup/A1000D1000/index.html

Read the above document and the Sun StorEdge?RAID Manager 6.22 Installation and Support Guide for Solaris (Part-No. 805-7756-10) and review at least the others.

If these documentions aren't suficient to get-you-going, I would strongly recommend that you get paid support from Sun or 3rd party.

Michael

Added the suggestion regarding hardware issues.

Message was edited by:

MAALATFT

MAALATFT at 2007-7-7 0:27:10 >

Hello,

there is no E220R is this my problem?,because the enterprise 220R is not supported by A1000

No, the Ultra 60/E220R are supported, but not listed in Sun StorEdge ?RAID Manager 6.22.1 Release Notes, Part-No. 805-7758-14. They are listed in InfoDoc 43483.

Please use the trouble-shooting tips from the documentation and my previous posts.

I (and everybody else) is unable to provide further assistance, if don't proceed in the correct order.

If you open a service case, you have to provide certain information and follow the instructions. If you don't, the issues can't be resolved.

You neither gave suficient informations nor followed the trouble-shooting tips.

Don't go to the next step, if an error occurs (it's upto you to decide, what an severe error is and what is only expected behaviour).

1. Get the A1000 detected at the ok-prompt. Don't boot until the device is listed with probe-scsi-all.

2. Boot the system into single-user mode with verbose messages and reconfiguration (boot -srv).

Watch carefully for errors.

3. Even without RaidManager the A1000 should be visible in format.

4. Check if the OBP up-to-date and required patches are installed.

5. The latest RaidManager won't cooperate with your A1000 until the firmware has been updated. Follow the steps in the documentation.

6. Even with earlier versions than Solaris 10, I would strongly recommend to make the modifications as described in my previous post. This results in much faster boot.

Michael

Some clarification and trouble-shooting procedure.

Message was edited by:

MAALATFT at 2007-7-7 0:27:10 >