3PAR node rescue and SPI Image

[555-denoh]

Hello everyone.

Could you please tell me about 3PAR7200?

We are currently verifying node rescue with 3PAR7200.
We think that the SPI images should be the same.

I think it is best to match the SPI of the node performing node rescue with the SPI of the cluster node.

Please tell me how to update the 3PAR 7200 node's SPI Image 2.4.0 to SPI Image 4.8.38.
I think it can be done with "whack>flash", but what do you think?

Thanks regard.

[veeyarvi]

Hi 555-denoh,

Normally SPI image has no effect on the node rescue. What is the actual issue you face during the rescue? Could you please share the exact error you see on the console?

Regards,

Veeyaarvi

[555-denoh]

Hello, 

Thank you for your message.
I will send you the results of the execution.

1.ide erase exected

Whack>i ahci init
Disk 0: AHCI0:
Model: SanDisk PU1000064GBATSSD Serial: **Confidential info erased** Rev: CS.51.00
Capacity: 64.023 GB (125045424 blks) SSD
WWN: 5001b449aecd
Interface: ATA-9 PIO mode: 0 Queue depth: 31 ECC bytes 0
Write cache ON
SMART ON
Power saving OFF
Whack>ie de erase　
ide commands unsupported on Sandy Bridge platforms
passing command to ahci...
Warning!! Security erase can take over an hour
Erase all AHCI0 drive data -- Are you sure? (Y/N) Yes
Erasing
.
Whack>reset　
Resetting...
Trapped IO 0xcf9 write of 0xa
<omitted>

2.Node rescue execution

--> Skipping: [45] Table Execution Summary
+-----------------------------------------------------------------------------+
HP SPI Image 2.4.0. Release version. 11:23:38 Apr 16 2013.
+-----------------------------------------------------------------------------+
| CPU 1 x 1.80 GHz Sandy Bridge quad core HT dual
| Control Cache Size 7.100 GB (cpu mem type 11) CL9
| Pair0 DIMM0 (J0155): 8192 MB CL5/10
| Data Cache Size 4 GB CL4/6
| Set0 DC DIMM 0.0.0 (J0300): 2048 MB CL4/6
| Set0 DC DIMM 0.1.0 (J0301): 2048 MB CL4/6
| Slot ID 1 [2 Node HP 3PAR 7200 Centerplane]
| FPGA Eos v0.e
| SATA0 Disk SanDisk PU1000064GBATSSD 131587401834 64 GB
| PCI Slot 0 LSI-SAS 9205-8e
| PCI Slot 1 Emulex LPe12002 VM80175440 2-port
| Board 920-200040.A8 FXN 2013/38/Tue 00196346 a4:5d:36:2d:1c:dc
| Cluster serial **confidential info erased** a4:5d:36:2d:1c:dd
| Board reset reason PCI_RESET
| Current Time 2001-01-24 00:14:30 (UTC-4)
+-----------------------------------------------------------------------------+
Booting from device 0, AHCI0Invalid boot sector.
Use "boot net install" to correct this.

*** Fatal error: Code 39, sub-code 0x10 (0). 
PROM detected error log time base problem. Verify RTC date
Log newest is "2016-11-26 20:23:19" RTC says it's "2001-01-24 00:14:31"
AHCI0 MBR does not have a valid partition table
Disabled FSBC boot watchdog.

*** Fatal error: Code 39, sub-code 0x10 (0).

*** Fatal error: Code 39, sub-code 0x10 (0).
Disabled FSBC boot watchdog.
Whack>

===============================
Don't go any further than this.
Node doesn't recognize the storage.
Any advice would be appreciated.

Thanks Regard.

[veeyarvi]

Hi Hajime,

The error code "Fatal error: Code 39, sub-code 0x10 (0)." indicates OS start up error or invalid boot sector. 

Is there a new IDE drive installed in this node? If yes, the node rescue was not successful or the drive (or related hardware is faulty). The disk does not have a valid partition table.

Did you try 'boot net install' from whack? Or you can try a manual node rescue from the CLI using 'startnoderescue <node ID>' and check.

Regards,
Veeyaarvi

[555-denoh]

Hello Veeyaarvi-san.

This is Hajime Shimaya.
Thank you for your support.

I tried 'boot net install' from whack and startnode rescue  on Two nodes.

I am initializing the SSD with Whack>ide erase.
Is this correct?
Are there any other commands?

Thanks Regard.

[555-denoh]

Hello,
This is Hajime Shimaya
I performed the following operations:

Whack>unset sys_serial
Whack>ahci init
Disk 0: AHCI0:
　　Model: SanDisk SSD U100 64GB Serial: 123612401529 Rev: CS.51.00
　　Capacity: 64.023 GB (125045424 blks) SSD
　　　WWN: 5001b4403412
　　Interface: ATA-9 PIO mode: 0 Queue depth: 31 ECC bytes 0
　　　Write cache ON
　　　SMART ON
　　　Power saving OFF
Whack>cb 2000 0 len=200;ide write 2000 0 1
ide commands unsupported on Sandy Bridge platforms
passing command to ahci...
Whack>reset

*******Run "Startnoderescue -node 1"  from the 3PAR side. *******

+-----------------------------------------------------------------------------+
HP SPI Image 1.2.9. Release version. 10:55:06 Nov 16 2012.
+-----------------------------------------------------------------------------+
<Omitted>
|
| Cluster serial 1613873 84:34:97:66:04:05
| Board reset reason PCI_RESET
| Current Time 2024-08-29 21:09:50 (UTC-4)
+-----------------------------------------------------------------------------+
Booting from device 0, AHCI0Invalid boot sector.
Use "boot net install" to correct this.

*** Fatal error: Code 39, sub-code 0x10 (0).
AHCI0 MBR does not have a valid partition table
Disabled FSBC boot watchdog.
Whack>

When the SSD that had experienced Fatal Error 39 was inserted into the original Node SPI4.8.38, Node Rescue was successful.
When the original SSD that was working was inserted into this Node with SPI1.2.9, it stopped with Fatal Error 39 and Node Rescue failed.

Is SPI1.2.9 Node too old to use?

Thankd Redard.

[veeyarvi]

Hi Shimaya,

The issue should be with the node itself if the same internal disk working on another node. The SPI image has little dpendency on the node rescue.

Regards,

Veeyaarvi

[555-denoh · Accepted solution]

Hello .
This is Hajime Shimaya
Thank you for your advice.　
We have solved this problem.

<>Purpose>
We need to execute 3PAR OS on a New 2 nodes for The inspection 

<Action plan>
I installed the 3PAR OS 3.2.2 (eMU4) SSD of equipment 7200 in a new node.
I boot up it with that combination. After booting it, I will try to upgrade the SPI with the upgradecage command.

<Progress>
1. Insert the 3PAR OS 3.2.2 (eMU4) SSD of equipment 7200 into the two new nodes .
2. Booted normally, and I found the update from SPI 1.2.9 to SPI 4.8.38 when the OS Booting up.
 This is a behavior I hadn't noticed before. 
===========Console Message================
Currently in the main SPI flash memory:  [5632 KB]
  HP SPI Image 1.2.9.  Release version. 10:55:06 Nov 16 2012.  
SPI image loaded:  [8192 KB] 
  HP SPI Image 4.8.38.  Release version. 16:57:04 Jul 17 2015.
    will be written to the main flash memory.
Flash Update Task Listing:
  #: Dst        Src        Length     Flags
  -- ---------- ---------- ---------- -------------------------
  0: 0x00000000 0x00000000 0x00010000 (0x00)
  1: 0x00280000 0x00280000 0x00580000 (0x88) BIOS Reboot
Erasing SPI at: 0x00000000; Length: 0x00010000
Verifying Erase at: 0x00000000; Length: 0x00010000
Writing SPI at: 0x00000000; Length: 0x00010000
Verifying SPI at: 0x00000000; Length: 0x00010000
Erasing SPI at: 0x00280000; Length: 0x00580000
Erasing blocks: [           ]...........]  
Verifying Erase at: 0x00280000; Length: 0x00580000
Verifying erase:  [           ]...........]  
Writing SPI at: 0x00280000; Length: 0x00580000
Writing bytes:  [           ]...........]  
Verifying SPI at: 0x00280000; Length: 0x00580000
Verifying bytes:  [           ]...........]  
Flash updates complete.  You may now reboot
===========Console Message================

3. New Nodes boot up Login prompt, but I can't login in with 3paradm/3pardata.
4. When I access from SP 7, 1
===========Console Message================
Could not open connection to CLI server process on "192.168.40.72":

              Socket connection timed out to port 2550 on 192.168.40.72
System name or port may be invalid, or the process may not have started yet or may have failed.
===========Console Message================
but I can't login 
The node port changed, so that's where the error occurred. Probably because the MAC address changed.

5.I couldn't shutdown command, so I pulled out the node and stopped 3PAR Node.
6.Returned the SSD to the original.
7. Started equipment 7200 with the original configuration.
8. Performed node  rescue on two new nodes with SPI 4.8.38.
   Both were successful. The inspection is now complete.

Thanks Regard.

[Sunitha_Mod]

@555-denoh 

Hello Hajime Shimaya,

That's Awesome! 

We are extremely glad to know the issue has been resolved and we appreciate you for keeping us updated.