Booting sparc T4 with just one of the mirrored rpool disks panics

Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 2933 bytes --]

Hi, hope someone knows what's happening here.
I have a T4 with XStreamOS sparc booting from a mirrored rpool, no problem here.
I then remove one of the disks of the mirror, and reboot.
During boot, system starts than panics with this output:
NOTICE: Performing full ZFS device scan!
strplumb: failed to initialize drv/dld
panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
Warning - stack not written to the dumpbuf
000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6013d74e188, 7bff9e80, 7bfe1990, 7b0, 109fa00, 0)
  %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
  %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 7b0, ab83c2f680, 0, 0)
  %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
  %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
000000000180b880 consconfig:consconfig+4 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000198f000 000006013d74e2c8 0000000001842d90 000003001827e000
  %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003001835c1e0
000000000180b930 unix:stubs_common_code+70 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000180b1f9 000000000180b2d1 0000003ffa000000 000000000180b9f0
  %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
  %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
  %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
skipping system dump - no dump device configured
rebooting...

This is the eeprom config output:

ttya-rts-dtr-off=false
ttya-ignore-cd=true
keyboard-layout: data not available.
reboot-command: data not available.
security-mode=none
security-password: data not available.
security-#badlogins=0
verbosity=min
diag-switch?=false
local-mac-address?=true
fcode-debug?=false
scsi-initiator-id=7
oem-logo: data not available.
oem-logo?=false
oem-banner: data not available.
oem-banner?=false
ansi-terminal?=true
screen-#columns=80
screen-#rows=34
ttya-mode=9600,8,n,1,-
output-device=virtual-console
input-device=virtual-console
auto-boot-on-error?=false
load-base=16384
auto-boot?=false
os-root-device: data not available.
network-boot-arguments: data not available.
boot-command=boot
boot-file: data not available.
boot-device=disk net
multipath-boot?=false
boot-device-index=0
use-nvramrc?=false
nvramrc: data not available.
error-reset-recovery=boot
What's wrong here? Looks like a console problem when mirror is not ok.

Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 


[-- Attachment #1.2: Type: text/html, Size: 8265 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 4181 bytes --]

Still on this problem...looks like the problem is here in cons_build_upper_layer of usr/src/uts/common/io/consconfig_dacf.c :

        /* open the console keyboard device.  will never be closed */
        if (ldi_open_by_name(CONSKBD_PATH, FREAD|FWRITE|FNOCTTY,
            kcred, &sp->conskbd_lh, sp->cons_li) != 0) {
                panic("consconfig: unable to open conskbd device");
                /*NOTREACHED*/
        }
 
Why this doesn't happen while the mirror boot is fine, and just panic if one of the two disks is missing?
I also tried to change the eeprom settings output-device and input-device to ttya (default is virtual-console).
Nothing changed...

Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Gabriele Bulfon via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 30 agosto 2023 15.31.35 CEST
Oggetto: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics



Hi, hope someone knows what's happening here.
I have a T4 with XStreamOS sparc booting from a mirrored rpool, no problem here.
I then remove one of the disks of the mirror, and reboot.
During boot, system starts than panics with this output:
NOTICE: Performing full ZFS device scan!
strplumb: failed to initialize drv/dld
panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
Warning - stack not written to the dumpbuf
000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6013d74e188, 7bff9e80, 7bfe1990, 7b0, 109fa00, 0)
  %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
  %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 7b0, ab83c2f680, 0, 0)
  %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
  %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
000000000180b880 consconfig:consconfig+4 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000198f000 000006013d74e2c8 0000000001842d90 000003001827e000
  %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003001835c1e0
000000000180b930 unix:stubs_common_code+70 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000180b1f9 000000000180b2d1 0000003ffa000000 000000000180b9f0
  %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
  %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
  %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
skipping system dump - no dump device configured
rebooting...

This is the eeprom config output:

ttya-rts-dtr-off=false
ttya-ignore-cd=true
keyboard-layout: data not available.
reboot-command: data not available.
security-mode=none
security-password: data not available.
security-#badlogins=0
verbosity=min
diag-switch?=false
local-mac-address?=true
fcode-debug?=false
scsi-initiator-id=7
oem-logo: data not available.
oem-logo?=false
oem-banner: data not available.
oem-banner?=false
ansi-terminal?=true
screen-#columns=80
screen-#rows=34
ttya-mode=9600,8,n,1,-
output-device=virtual-console
input-device=virtual-console
auto-boot-on-error?=false
load-base=16384
auto-boot?=false
os-root-device: data not available.
network-boot-arguments: data not available.
boot-command=boot
boot-file: data not available.
boot-device=disk net
multipath-boot?=false
boot-device-index=0
use-nvramrc?=false
nvramrc: data not available.
error-reset-recovery=boot
What's wrong here? Looks like a console problem when mirror is not ok.

Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 11915 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Toomas Soome]

[-- Attachment #1: Type: text/plain, Size: 5026 bytes --]



> On 5. Sep 2023, at 13:25, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
> 
> Still on this problem...looks like the problem is here in cons_build_upper_layer of usr/src/uts/common/io/consconfig_dacf.c :
> 
>         /* open the console keyboard device.  will never be closed */
>         if (ldi_open_by_name(CONSKBD_PATH, FREAD|FWRITE|FNOCTTY,
>             kcred, &sp->conskbd_lh, sp->cons_li) != 0) {
>                 panic("consconfig: unable to open conskbd device");
>                 /*NOTREACHED*/
>         }
>  
> Why this doesn't happen while the mirror boot is fine, and just panic if one of the two disks is missing?
> I also tried to change the eeprom settings output-device and input-device to ttya (default is virtual-console).
> Nothing changed...
> 
> Gabriele


Please note, the sparc zfs boot block does not really switch to alternate disk if read is failing, you need to boot from good disk (boot diskX).

rgds,
toomas

>  
>  
> Sonicle S.r.l. : http://www.sonicle.com <https://www.sonicle.com/>
> Music: http://www.gabrielebulfon.com <http://www.gabrielebulfon.com/>
> eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
>  
>  
> 
> 
> Da: Gabriele Bulfon via illumos-developer <developer@lists.illumos.org>
> A: illumos-developer <developer@lists.illumos.org>
> Data: 30 agosto 2023 15.31.35 CEST
> Oggetto: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics
> 
> 
> Hi, hope someone knows what's happening here.
> I have a T4 with XStreamOS sparc booting from a mirrored rpool, no problem here.
> I then remove one of the disks of the mirror, and reboot.
> During boot, system starts than panics with this output:
> NOTICE: Performing full ZFS device scan!
> strplumb: failed to initialize drv/dld
> 
> panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
> 
> Warning - stack not written to the dumpbuf
> 000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6013d74e188, 7bff9e80, 7bfe1990, 7b0, 109fa00, 0)
>   %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
>   %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
> 000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 7b0, ab83c2f680, 0, 0)
>   %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
>   %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
> 000000000180b880 consconfig:consconfig+4 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
>   %l0-3: 000000000198f000 000006013d74e2c8 0000000001842d90 000003001827e000
>   %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003001835c1e0
> 000000000180b930 unix:stubs_common_code+70 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
>   %l0-3: 000000000180b1f9 000000000180b2d1 0000003ffa000000 000000000180b9f0
>   %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
> 000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
>   %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
>   %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
> 
> skipping system dump - no dump device configured
> rebooting...
> 
> This is the eeprom config output:
> 
> ttya-rts-dtr-off=false
> ttya-ignore-cd=true
> keyboard-layout: data not available.
> reboot-command: data not available.
> security-mode=none
> security-password: data not available.
> security-#badlogins=0
> verbosity=min
> diag-switch?=false
> local-mac-address?=true
> fcode-debug?=false
> scsi-initiator-id=7
> oem-logo: data not available.
> oem-logo?=false
> oem-banner: data not available.
> oem-banner?=false
> ansi-terminal?=true
> screen-#columns=80
> screen-#rows=34
> ttya-mode=9600,8,n,1,-
> output-device=virtual-console
> input-device=virtual-console
> auto-boot-on-error?=false
> load-base=16384
> auto-boot?=false
> os-root-device: data not available.
> network-boot-arguments: data not available.
> boot-command=boot
> boot-file: data not available.
> boot-device=disk net
> multipath-boot?=false
> boot-device-index=0
> use-nvramrc?=false
> nvramrc: data not available.
> error-reset-recovery=boot
> 
> What's wrong here? Looks like a console problem when mirror is not ok.
> 
> Gabriele
>  
>  
> Sonicle S.r.l. : http://www.sonicle.com <https://www.sonicle.com/>
> Music: http://www.gabrielebulfon.com <http://www.gabrielebulfon.com/>
> eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
>  
> illumos <https://illumos.topicbox.com/latest> / illumos-developer / see discussions <https://illumos.topicbox.com/groups/developer> + participants <https://illumos.topicbox.com/groups/developer/members> + delivery options <https://illumos.topicbox.com/groups/developer/subscription>Permalink <https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-M50dc90a64b9e24e14f9c2c90>

[-- Attachment #2: Type: text/html, Size: 12386 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 14697 bytes --]

Thanks Toomas, tried what you suggested, same problem, here is the full stack from HW boot.
I can't understand why this happens only with a single drive of a mirror, not with the full mirror operating.
Someone at FB Sun group suggested that on sparc is not ok to detach a disk physically, you need to remove it from the pool before detaching, but:
- if I detach it from the pool, that disk will be unusable (marked as no more part of a mirror, no boot), unless there is another way to remove it from the pool and keep it untouched.
- that is exactly what I was trying to simulate: a physical error. That same guy said that only on system detected error that disk can be removed, and the running disk will be correctly bootable....strange to me!

What I'm trying to achive is to simulate failure of boot on one of the disk and be sure the original machine will correctly boot with one disk, also that if I want to create a third mirror disk, I can remove that third disk and reuse it on another T4.

-> start /SYS
Are you sure you want to start /SYS (y/n)? y
Starting /SYS
-> start /SP/console
Are you sure you want to start /SP/console (y/n)? y
Serial console started.  To stop, type #.
2002-06-19 01:52:36  0:0:0> NOTICE:  Initializing TPM with:
                        tpm_enable = false
                        tpm_activate = false
                        tpm_forceclear = false
2002-06-19 01:52:36  0:0:0> NOTICE:  TPM found: Ver 1.2, Rev 1.2, SpecLevel 2, e    rrataRev 0, VendorId 'IFX'
2002-06-19 01:52:37  0:0:0> NOTICE:  TPM initialized successfully. Current state     is: disabled
2002-06-19 01:52:37  0:0:0> NOTICE:  Serial#:     000000000000002a.015b58c07e2da    f25
2002-06-19 01:52:37  0:0:0> NOTICE:  Version:     003e003013030607
2002-06-19 01:52:37  0:0:0> NOTICE:  T4 Revision: 1.3
2002-06-19 01:52:40  0:0:0> NOTICE:  MCU0: Memory Capacity is 16GB
2002-06-19 01:52:40  0:0:0> NOTICE:  MCU1: Memory Capacity is 16GB
2002-06-19 01:52:43  0:0:0> NOTICE:  Usable strands: ffffffffffffffff
2002-06-19 01:52:43  0:0:0> NOTICE:  System memory capacity is 32GB
2002-06-19 01:52:46  0:0:0> NOTICE:  Clocks: CMP: 2848 MHz DRAM: 533 MHz (6.4 Gb    ps) CL: 1466 MHz (8.8 Gbps)
2002-06-19 01:52:46  0:0:0> NOTICE:  Initializing TSR Hoovers
2002-06-19 01:52:47  0:0:0> NOTICE:  Initializing FSR Hoovers
2002-06-19 01:52:49  0:0:0> NOTICE:  Initializing MCU 0 serdes
2002-06-19 01:52:49  0:0:0> NOTICE:  Initializing MCU 1 serdes
2002-06-19 01:52:49  0:0:0> NOTICE:  Updating Config Information for Guest Manag    er
2002-06-19 01:52:52  0:0:0> NOTICE:  Enabling caches
2002-06-19 01:52:52  0:0:0> NOTICE:  L3 Banks Enabled: ff
2002-06-19 01:52:54  0:0:0> NOTICE:  Initializing MCU 0
2002-06-19 01:52:54  0:0:0> NOTICE:  Initializing MCU 1
2002-06-19 01:53:03  0:0:0> NOTICE:  SMI Channel 0, SB Mapping 0 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:03  0:0:0> NOTICE:  SMI Channel 0, SB Mapping 1 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:03  0:0:0> NOTICE:  SMI Channel 1, SB Mapping 0 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:03  0:0:0> NOTICE:  SMI Channel 1, SB Mapping 1 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:18  0:0:0> NOTICE:  SMI Channel 0, SB Mapping 0 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:18  0:0:0> NOTICE:  SMI Channel 0, SB Mapping 1 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:18  0:0:0> NOTICE:  SMI Channel 1, SB Mapping 0 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:18  0:0:0> NOTICE:  SMI Channel 1, SB Mapping 1 -- ERRCNT:    0    x0      LNERR:  0x0
2002-06-19 01:53:36  0:0:0> NOTICE:  L3 Banks Enabled: ff
2002-06-19 01:53:41  0:0:0> NOTICE:  Copying code to memory
2002-06-19 01:53:43  0:0:0> NOTICE:  Copy to memory done
2002-06-19 01:53:44  0:0:0> NOTICE:  Running from memory
2002-06-19 01:53:44  0:0:0> NOTICE:  Starting available strands
2002-06-19 01:53:44  0:0:0> NOTICE:  Active strands:    ffffffffffffffff
2002-06-19 01:53:44  0:0:0> NOTICE:  Configuring MDs
2002-06-19 01:53:44  0:0:0> NOTICE:  Loading PRI template
2002-06-19 01:53:44  0:0:0> NOTICE:  Configuring PRI
2002-06-19 01:53:45  0:0:0> NOTICE:  Product serial number: 1231BDYD6E
2002-06-19 01:53:45  0:0:0> NOTICE:  Product part number: 30972757+1+1
2002-06-19 01:53:46  0:0:0> NOTICE:  Memory on node 0 isn't power manageable
2002-06-19 01:53:46  0:0:0> NOTICE:  Storing PRI to memory
2002-06-19 01:53:46  0:0:0> NOTICE:  Booting config = factory-default
2002-06-19 01:53:46  0:0:0> NOTICE:  Configuring Guest MD
2002-06-19 01:53:46  0:0:0> NOTICE:  Storing Guest MD to Memory
2002-06-19 01:53:46  0:0:0> NOTICE:  Configuring HV MD
2002-06-19 01:53:46  0:0:0> NOTICE:  Storing HV MD to Memory
2002-06-19 01:53:47  0:0:0> NOTICE:  Storing Machine Descriptors to data flash
2002-06-19 01:53:47  0:0:0> NOTICE:  Updating Config Information for Guest Manag    er
2002-06-19 01:53:47  0:0:0> NOTICE:  Jumping to hypervisor
[CPU 00:00:0] Hypervisor version: @(#)Hypervisor 1.15.16 2018/11/28 07:41
NOTICE: Entering OpenBoot.
NOTICE: Fetching Guest MD from HV.
NOTICE: Starting additional cpus.
NOTICE: Initializing LDC services.
NOTICE: Probing PCI devices.
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard
Copyright (c) 1998, 2018, Oracle and/or its affiliates. All rights reserved.
OpenBoot 4.38.16, 31.5000 GB memory available, Serial #101615306.
Ethernet address 0:10:e0:e:86:ca, Host ID: 860e86ca.

{0} ok boot disk0
Boot device: /pci@400/pci@1/pci@0/pci@4/scsi@0/disk@p0  File and args:
SunOS Release 5.11 Version SunriseOS Sparc 64-bit July 2023
SunriseOS is mantained and distributed by Sunrise (https://www.sunrise-it.net).
All other trademarks and copyrights are the property of their respective owners.
NOTICE: Performing full ZFS device scan!
strplumb: failed to initialize drv/dld
panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
Warning - stack not written to the dumpbuf
000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6003a512688, 7bff9e80, 7bfe1990, 100, fe1800, 0)
  %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
  %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 100, 1eedb3c340, 0, 0)
  %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
  %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
000000000180b880 consconfig:consconfig+4 (1b70f0d4d1, e3ed6, 10031304, 0, fe5700, 0)
  %l0-3: 000000000198f000 000006003a5127c8 0000000001842d90 000003000a222000
  %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003000a1d41e0
000000000180b930 unix:stubs_common_code+70 (1b70f0d4d1, e3ed6, 10031304, 0, fe5700, 0)
  %l0-3: 000000000180b1f9 000000000180b2d1 00000007fa000000 000000000180b9f0
  %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
  %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
  %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
skipping system dump - no dump device configured
rebooting...
NOTICE: Entering OpenBoot.
NOTICE: Fetching Guest MD from HV.
NOTICE: Starting additional cpus.
NOTICE: Initializing LDC services.
NOTICE: Probing PCI devices.
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard
Copyright (c) 1998, 2018, Oracle and/or its affiliates. All rights reserved.
OpenBoot 4.38.16, 31.5000 GB memory available, Serial #101615306.
Ethernet address 0:10:e0:e:86:ca, Host ID: 860e86ca.

Boot device: disk  File and args:
SunOS Release 5.11 Version SunriseOS Sparc 64-bit July 2023
SunriseOS is mantained and distributed by Sunrise (https://www.sunrise-it.net).
All other trademarks and copyrights are the property of their respective owners.
NOTICE: Performing full ZFS device scan!
strplumb: failed to initialize drv/dld
panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
Warning - stack not written to the dumpbuf
000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6003a512688, 7bff9e80, 7bfe1990, 1c60, f18b00, 0)
  %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
  %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 1c60, 384b360500, 0, 0)
  %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
  %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
000000000180b880 consconfig:consconfig+4 (34ce5cf9f5, 44b25, 10031304, 0, f1c700, 0)
  %l0-3: 000000000198f000 000006003a5127c8 0000000001842d90 000003000a1da000
  %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003000a1b81e0
000000000180b930 unix:stubs_common_code+70 (34ce5cf9f5, 44b25, 10031304, 0, f1c700, 0)
  %l0-3: 000000000180b1f9 000000000180b2d1 00000007fa000000 000000000180b9f0
  %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
  %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
  %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
skipping system dump - no dump device configured
rebooting...

 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 12.28.15 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 13:25, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
Still on this problem...looks like the problem is here in cons_build_upper_layer of usr/src/uts/common/io/consconfig_dacf.c :

        /* open the console keyboard device.  will never be closed */
        if (ldi_open_by_name(CONSKBD_PATH, FREAD|FWRITE|FNOCTTY,
            kcred, &sp->conskbd_lh, sp->cons_li) != 0) {
                panic("consconfig: unable to open conskbd device");
                /*NOTREACHED*/
        }
 
Why this doesn't happen while the mirror boot is fine, and just panic if one of the two disks is missing?
I also tried to change the eeprom settings output-device and input-device to ttya (default is virtual-console).
Nothing changed...

Gabriele


 
 
Please note, the sparc zfs boot block does not really switch to alternate disk if read is failing, you need to boot from good disk (boot diskX).
 
rgds,
toomas

 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Gabriele Bulfon via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 30 agosto 2023 15.31.35 CEST
Oggetto: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics



Hi, hope someone knows what's happening here.
I have a T4 with XStreamOS sparc booting from a mirrored rpool, no problem here.
I then remove one of the disks of the mirror, and reboot.
During boot, system starts than panics with this output:
NOTICE: Performing full ZFS device scan!
strplumb: failed to initialize drv/dld
panic[cpu0]/thread=180e000: consconfig: unable to open conskbd device
Warning - stack not written to the dumpbuf
000000000180b6f0 consconfig_dacf:cons_build_upper_layer+230 (6013d74e188, 7bff9e80, 7bfe1990, 7b0, 109fa00, 0)
  %l0-3: 000000000198d000 0000000001a0c400 0000000000000002 00000000013016d8
  %l4-7: 0000000001945ff0 0000000000080000 0000000000000000 0000000000000001
000000000180b7d0 consconfig_dacf:dynamic_console_config+70 (198f190, 180e000, 7b0, ab83c2f680, 0, 0)
  %l0-3: 0000000001a0c400 0000000001999400 000000000193e000 000000000193e000
  %l4-7: 0000000000000000 000000000193e000 0000000001999400 0000000001998000
000000000180b880 consconfig:consconfig+4 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000198f000 000006013d74e2c8 0000000001842d90 000003001827e000
  %l4-7: 0000000000000010 0000000000000000 0000000000000002 000003001835c1e0
000000000180b930 unix:stubs_common_code+70 (a8072be474, a7598, 10031304, 0, 10a4200, 0)
  %l0-3: 000000000180b1f9 000000000180b2d1 0000003ffa000000 000000000180b9f0
  %l4-7: 0000000000000000 0000000001842da0 0000000000000000 0000000000000000
000000000180ba00 genunix:main+1b0 (180bb80, 180bec0, 0, 0, f02662a4, 0)
  %l0-3: 00000000018a4ec0 0000000000000001 0000000001997c00 000000000180bad0
  %l4-7: 000000000190bb68 00000000018a4ec0 0000000000000000 00000000f0800da0
skipping system dump - no dump device configured
rebooting...

This is the eeprom config output:

ttya-rts-dtr-off=false
ttya-ignore-cd=true
keyboard-layout: data not available.
reboot-command: data not available.
security-mode=none
security-password: data not available.
security-#badlogins=0
verbosity=min
diag-switch?=false
local-mac-address?=true
fcode-debug?=false
scsi-initiator-id=7
oem-logo: data not available.
oem-logo?=false
oem-banner: data not available.
oem-banner?=false
ansi-terminal?=true
screen-#columns=80
screen-#rows=34
ttya-mode=9600,8,n,1,-
output-device=virtual-console
input-device=virtual-console
auto-boot-on-error?=false
load-base=16384
auto-boot?=false
os-root-device: data not available.
network-boot-arguments: data not available.
boot-command=boot
boot-file: data not available.
boot-device=disk net
multipath-boot?=false
boot-device-index=0
use-nvramrc?=false
nvramrc: data not available.
error-reset-recovery=boot
What's wrong here? Looks like a console problem when mirror is not ok.

Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 





illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 25300 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Toomas Soome]

[-- Attachment #1: Type: text/plain, Size: 313 bytes --]



> On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
> 
> NOTICE: Finished PCI probing.
> ttya not found.
> ttya not found.
> 
> SPARC T4-1, No Keyboard
> 

thats interesting, if you get to ok prompt, does devalias command list ttya?

rgds,
toomas

[-- Attachment #2: Type: text/html, Size: 1358 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 923 bytes --]

That's a recent try (because of the console errors), changed from virtual to ttya, no difference...
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 14.17.14 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard



thats interesting, if you get to ok prompt, does devalias command list ttya?
 
rgds,
toomas
illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 3619 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gergő Doma]

[-- Attachment #1: Type: text/plain, Size: 1556 bytes --]

Is there anything interesting in the output of the "::zfs_dbgmsg" command?

Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> ezt
írta (időpont: 2023. szept. 5., K, 15:18):

> That's a recent try (because of the console errors), changed from virtual
> to ttya, no difference...
>
>
> *Sonicle S.r.l. *: http://www.sonicle.com <https://www.sonicle.com/>
> *Music: *http://www.gabrielebulfon.com
> *eXoplanets : *https://gabrielebulfon.bandcamp.com/album/exoplanets
>
>
> ------------------------------
>
>
> *Da:* Toomas Soome via illumos-developer <developer@lists.illumos.org>
> *A:* illumos-developer <developer@lists.illumos.org>
> *Data:* 5 settembre 2023 14.17.14 CEST
> *Oggetto:* Re: [developer] Booting sparc T4 with just one of the mirrored
> rpool disks panics
>
>
>
>
> On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <
> developer@lists.illumos.org> wrote:
>
> NOTICE: Finished PCI probing.
> ttya not found.
> ttya not found.
>
> SPARC T4-1, No Keyboard
>
>
> thats interesting, if you get to ok prompt, does devalias command list
> ttya?
>
> rgds,
> toomas
>
> *illumos <https://illumos.topicbox.com/latest>* / illumos-developer / see
> discussions <https://illumos.topicbox.com/groups/developer> + participants
> <https://illumos.topicbox.com/groups/developer/members> + delivery options
> <https://illumos.topicbox.com/groups/developer/subscription> Permalink
> <https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-Mbf8b94cbb25f7587261a527d>
>

[-- Attachment #2: Type: text/html, Size: 4206 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 1650 bytes --]

Thanks, is this a command that needs to be activated some way?
The mdb command says it's not a dcmd command.
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Gergő Doma <domag02@gmail.com>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 22.00.24 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics



Is there anything interesting in the output of the "::zfs_dbgmsg" command?

Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> ezt írta (időpont: 2023. szept. 5., K, 15:18):
That's a recent try (because of the console errors), changed from virtual to ttya, no difference...
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 14.17.14 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard




thats interesting, if you get to ok prompt, does devalias command list ttya?
 
rgds,
toomas



illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 6498 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Nikola M.]

[-- Attachment #1: Type: text/plain, Size: 1548 bytes --]

On 9/5/23 12:28, Toomas Soome via illumos-developer wrote:
>
> Please note, the sparc zfs boot block does not really switch to 
> alternate disk if read is failing, you need to boot from good disk 
> (boot diskX).
>
> rgds,
> toomas

Hello, I am just sharing to ask for possible check and confirmation.

Meaning, one needs to install OS on one drive, that makes it bootable, 
then add second drive in the mirror afterwards but doing
( 
https://docs.oracle.com/cd/E19120-01/open.solaris/819-2723/disksprep-103/index.html 
)
( 
https://dlc.openindiana.org/docs/osol/20090715/SAGDFS/html/disksprep-101.html 
)

prtvtoc /dev/rdsk/cwtxdysz | fmthard -s - /dev/rdsk/cwt${i}d0s2

zpool Attach new drive to create a mirror

and then " Install a Boot Block on a System Disk"
( 
https://docs.oracle.com/cd/E19120-01/open.solaris/819-2723/diskssadd-16103/index.html 
)
( 
https://dlc.openindiana.org/docs/osol/20090715/SAGDFS/html/diskssadd-53.html#diskssadd-43588 
)

# installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk 
/dev/rdsk/cwtxdys0

But is it using GRUB or illumos loader on XStreamOS SPARC?

It is saying in that old manual that installing boot loader is done 
after attaching second drive to the boot pool.

I haven't tried adding drive XStreamOS on SPARC yet, (Actually I have 
one single-disk opensolaris install on T2 SPARC, that I pulled second 
drive from, so I might try.)
Please confirm if it is the case how it should be done to be bootable 
even on mirrored boot disk failure.

Cheers,

[-- Attachment #2: Type: text/html, Size: 3041 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 2631 bytes --]

SPARC systems feature openboot, not grub or recent x86 illumos loader.
When you attach to create mirror, if the original disk was bootable recent illumos automatically makes new disk bootable too, AFAIK.
 
The problem to me here is not about boot: the system boots, you can see the kernel boot message, then karnel panics when it recognizes that the rpool mirror is broken, it probably tries to send some message to the console but than fails to open the console keyboard device..........and that doesn't happen while the mirror is sane......that's so strange, it completely breaks the possibility to boot the system with just one disk of the mirror.
I feel unsafe with this situation, if one of the disks fails for hw errors, I won't be able to boot the system anymore....

Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Nikola M. <minikola@gmail.com>
A: illumos-developer <developer@lists.illumos.org>
Data: 6 settembre 2023 10.16.07 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics



On 9/5/23 12:28, Toomas Soome via illumos-developer wrote:
Hello, I am just sharing to ask for possible check and confirmation. 

Meaning, one needs to install OS on one drive, that makes it bootable, then add second drive in the mirror afterwards but doing 
( https://docs.oracle.com/cd/E19120-01/open.solaris/819-2723/disksprep-103/index.html ) 
( https://dlc.openindiana.org/docs/osol/20090715/SAGDFS/html/disksprep-101.html ) 

prtvtoc /dev/rdsk/cwtxdysz | fmthard -s - /dev/rdsk/cwt${i}d0s2

zpool Attach new drive to create a mirror 

and then " Install a Boot Block on a System Disk" 
( https://docs.oracle.com/cd/E19120-01/open.solaris/819-2723/diskssadd-16103/index.html )
( https://dlc.openindiana.org/docs/osol/20090715/SAGDFS/html/diskssadd-53.html#diskssadd-43588 )

# installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk  /dev/rdsk/cwtxdys0 

But is it using GRUB or illumos loader on XStreamOS SPARC? 

It is saying in that old manual that installing boot loader is done after attaching second drive to the boot pool. 

I haven't tried adding drive XStreamOS on SPARC yet, (Actually I have one single-disk opensolaris install on T2 SPARC, that I pulled second drive from, so I might try.)
Please confirm if it is the case how it should be done to be bootable even on mirrored boot disk failure. 

Cheers, 
illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 5250 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 1223 bytes --]

Looks like a very similar problem to this, though here he was using SVM...:

https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
 
problem seems very similar.
I can't believe anyone ever tried to remove a disk from a mirror and boot back from the only one present, on SPARC, or mayb it's just about the T4...
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 14.17.14 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard



thats interesting, if you get to ok prompt, does devalias command list ttya?
 
rgds,
toomas
illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 4749 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Peter Tribble]

[-- Attachment #1: Type: text/plain, Size: 2416 bytes --]

On Wed, Sep 6, 2023 at 3:46 PM Gabriele Bulfon via illumos-developer <
developer@lists.illumos.org> wrote:

> Looks like a very similar problem to this, though here he was using SVM...:
>

I remember that one. That was a plain and simple bug in SVM (it required
you to have
a majority of metadevice dbs in order to boot, and exactly half isn't a
majority - we had
a process where if a disk failed we simply force deleted the metadevice dbs
on the
failed device and then everything's happy).

I don't know what the problem is here; all I would say that the specific
consconfig error
you're seeing doesn't mean anything - I think the system is panicing very
early, before
everything else has been properly initialized.


>
> https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
>
> problem seems very similar.
> I can't believe anyone ever tried to remove a disk from a mirror and boot
> back from the only one present, on SPARC, or mayb it's just about the T4...
>
> Gabriele
>
>
> *Sonicle S.r.l. *: http://www.sonicle.com <https://www.sonicle.com/>
> *Music: *http://www.gabrielebulfon.com
> *eXoplanets : *https://gabrielebulfon.bandcamp.com/album/exoplanets
>
>
> ------------------------------
>
>
> *Da:* Toomas Soome via illumos-developer <developer@lists.illumos.org>
> *A:* illumos-developer <developer@lists.illumos.org>
> *Data:* 5 settembre 2023 14.17.14 CEST
> *Oggetto:* Re: [developer] Booting sparc T4 with just one of the mirrored
> rpool disks panics
>
>
>
>
> On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <
> developer@lists.illumos.org> wrote:
>
> NOTICE: Finished PCI probing.
> ttya not found.
> ttya not found.
>
> SPARC T4-1, No Keyboard
>
>
> thats interesting, if you get to ok prompt, does devalias command list
> ttya?
>
> rgds,
> toomas
>
> *illumos <https://illumos.topicbox.com/latest>* / illumos-developer / see
> discussions <https://illumos.topicbox.com/groups/developer> + participants
> <https://illumos.topicbox.com/groups/developer/members> + delivery options
> <https://illumos.topicbox.com/groups/developer/subscription> Permalink
> <https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-Mc55d1de224143eb9e4889dc3>
>


-- 
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/

[-- Attachment #2: Type: text/html, Size: 6053 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Toomas Soome]

[-- Attachment #1: Type: text/plain, Size: 3123 bytes --]



> On 6. Sep 2023, at 18:14, Peter Tribble <peter.tribble@gmail.com> wrote:
> 
> On Wed, Sep 6, 2023 at 3:46 PM Gabriele Bulfon via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>> wrote:
>> Looks like a very similar problem to this, though here he was using SVM...:
> 
> I remember that one. That was a plain and simple bug in SVM (it required you to have
> a majority of metadevice dbs in order to boot, and exactly half isn't a majority - we had
> a process where if a disk failed we simply force deleted the metadevice dbs on the
> failed device and then everything's happy).


it was not bug but documented feature and the reason was about majority election mechanism used to decide if configuration is usable. So to survive, you needed to have 50% of metadb available and to boot, you needed 50% + 1. So with 2 disks setup, you had to split the number of metadb copies between 2 disks, and when losing one disk, before shutdown you had to remove instance(s) from failing disks to be able to boot:)

rgds,
toomas

> 
> I don't know what the problem is here; all I would say that the specific consconfig error
> you're seeing doesn't mean anything - I think the system is panicing very early, before
> everything else has been properly initialized.
>  
>> https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
>>  
>> problem seems very similar.
>> I can't believe anyone ever tried to remove a disk from a mirror and boot back from the only one present, on SPARC, or mayb it's just about the T4...
>>  
>> Gabriele
>>  
>>  
>> Sonicle S.r.l. : http://www.sonicle.com <https://www.sonicle.com/>
>> Music: http://www.gabrielebulfon.com <http://www.gabrielebulfon.com/>
>> eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
>>  
>>  
>> 
>> 
>> Da: Toomas Soome via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>>
>> A: illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>>
>> Data: 5 settembre 2023 14.17.14 CEST
>> Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics
>> 
>> 
>> 
>> 
>> On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>> wrote:
>> NOTICE: Finished PCI probing.
>> ttya not found.
>> ttya not found.
>> 
>> SPARC T4-1, No Keyboard
>> 
>> 
>> 
>> thats interesting, if you get to ok prompt, does devalias command list ttya?
>>  
>> rgds,
>> toomas
> 
> 
> -- 
> -Peter Tribble
> http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
> illumos <https://illumos.topicbox.com/latest> / illumos-developer / see discussions <https://illumos.topicbox.com/groups/developer> + participants <https://illumos.topicbox.com/groups/developer/members> + delivery options <https://illumos.topicbox.com/groups/developer/subscription>Permalink <https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-Mfa9d5a11973b347a089e1f1a>

[-- Attachment #2: Type: text/html, Size: 8832 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 3616 bytes --]

I bet it is the same reason....that "NOTICE: Performing full ZFS device scan!" never happened on x86 in the same situation.
That means I have to remove the disk from the mirror when physically removed or failed.
 
Probelm is, if I detach the disk, that disk won't be anymore usable alone for a second machine, so I will detach it from the mirror after it has been marked as degraded.
That removed disk, though, won't be able to boot alone on another machine, I need some way to transform it into a single disk pool.
Maybe I can import it on another running system, detach the missing disk and export it before booting it on another system?
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 6 settembre 2023 18.17.15 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 6. Sep 2023, at 18:14, Peter Tribble <peter.tribble@gmail.com> wrote:
On Wed, Sep 6, 2023 at 3:46 PM Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
Looks like a very similar problem to this, though here he was using SVM...:

 
I remember that one. That was a plain and simple bug in SVM (it required you to have
a majority of metadevice dbs in order to boot, and exactly half isn't a majority - we had
a process where if a disk failed we simply force deleted the metadevice dbs on the
failed device and then everything's happy).



 
 
it was not bug but documented feature and the reason was about majority election mechanism used to decide if configuration is usable. So to survive, you needed to have 50% of metadb available and to boot, you needed 50% + 1. So with 2 disks setup, you had to split the number of metadb copies between 2 disks, and when losing one disk, before shutdown you had to remove instance(s) from failing disks to be able to boot:)
 
rgds,
toomas

 
I don't know what the problem is here; all I would say that the specific consconfig error
you're seeing doesn't mean anything - I think the system is panicing very early, before
everything else has been properly initialized.
 
https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
 
problem seems very similar.
I can't believe anyone ever tried to remove a disk from a mirror and boot back from the only one present, on SPARC, or mayb it's just about the T4...
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 14.17.14 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard
 



thats interesting, if you get to ok prompt, does devalias command list ttya?
 
rgds,
toomas



-- 
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/




illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 11393 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Toomas Soome]

[-- Attachment #1: Type: text/plain, Size: 5680 bytes --]



> On 7. Sep 2023, at 09:54, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
> 
> I bet it is the same reason....that "NOTICE: Performing full ZFS device scan!" never happened on x86 in the same situation.
> That means I have to remove the disk from the mirror when physically removed or failed.
>  

ZFS is managing configurations differently (we have 4 copies of labels per disk), so that assumption does not hold there.

If you want to re-use half of the mirror in other system, then you would need to not just pull the disk out, but use zpool split and export it before disk removal.

Why the “broken mirror” fails to boot on SPARC — the one possible issue may be, the zfs boot block fails to try alternate block address when first one fails. In case of mirror, we have data stored as multiple copies and if the zfs reader will try missing copy first, it may be it fails to get to next one. But that is just an guess about what might be happening there (zfs reader is written in forth in case of SPARC). Of course, this is something we could improve, but since we have announced, we will not support SPARC… 

So, for workaround, this situation would mean that if you have broken disk in boot pool, you can use hot replace if hw does support it (zpool offline the disk, cfgadm -c unconfigure disk, replace it, then cfgadm -c configure, zpool replace). Or if hot replace is not possible and boot is failing, then boot from alternate media, import and resilver the pool and that should fix the pool.

rgds,
toomas


> Probelm is, if I detach the disk, that disk won't be anymore usable alone for a second machine, so I will detach it from the mirror after it has been marked as degraded.
> That removed disk, though, won't be able to boot alone on another machine, I need some way to transform it into a single disk pool.
> Maybe I can import it on another running system, detach the missing disk and export it before booting it on another system?
>  
> Gabriele
>  
>  
> Sonicle S.r.l. : http://www.sonicle.com <https://www.sonicle.com/>
> Music: http://www.gabrielebulfon.com <http://www.gabrielebulfon.com/>
> eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
>  
>  
> 
> 
> Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
> A: illumos-developer <developer@lists.illumos.org>
> Data: 6 settembre 2023 18.17.15 CEST
> Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics
> 
> 
> 
> 
> On 6. Sep 2023, at 18:14, Peter Tribble <peter.tribble@gmail.com> wrote:
> On Wed, Sep 6, 2023 at 3:46 PM Gabriele Bulfon via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>> wrote:
>> Looks like a very similar problem to this, though here he was using SVM...:
>  
> I remember that one. That was a plain and simple bug in SVM (it required you to have
> a majority of metadevice dbs in order to boot, and exactly half isn't a majority - we had
> a process where if a disk failed we simply force deleted the metadevice dbs on the
> failed device and then everything's happy).
>  
>  
> it was not bug but documented feature and the reason was about majority election mechanism used to decide if configuration is usable. So to survive, you needed to have 50% of metadb available and to boot, you needed 50% + 1. So with 2 disks setup, you had to split the number of metadb copies between 2 disks, and when losing one disk, before shutdown you had to remove instance(s) from failing disks to be able to boot:)
>  
> rgds,
> toomas
> 
>  
> I don't know what the problem is here; all I would say that the specific consconfig error
> you're seeing doesn't mean anything - I think the system is panicing very early, before
> everything else has been properly initialized.
>  
>> https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
>>  
>> problem seems very similar.
>> I can't believe anyone ever tried to remove a disk from a mirror and boot back from the only one present, on SPARC, or mayb it's just about the T4...
>>  
>> Gabriele
>>  
>>  
>> Sonicle S.r.l. : http://www.sonicle.com <https://www.sonicle.com/>
>> Music: http://www.gabrielebulfon.com <http://www.gabrielebulfon.com/>
>> eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
>>  
>>  
>> 
>> 
>> Da: Toomas Soome via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>>
>> A: illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>>
>> Data: 5 settembre 2023 14.17.14 CEST
>> Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics
>> 
>> 
>> 
>> 
>> On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org <mailto:developer@lists.illumos.org>> wrote:
>> NOTICE: Finished PCI probing.
>> ttya not found.
>> ttya not found.
>> 
>> SPARC T4-1, No Keyboard
>> 
>>  
>> 
>> thats interesting, if you get to ok prompt, does devalias command list ttya?
>>  
>> rgds,
>> toomas
> 
> 
> -- 
> -Peter Tribble
> http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
> 
> illumos <https://illumos.topicbox.com/latest> / illumos-developer / see discussions <https://illumos.topicbox.com/groups/developer> + participants <https://illumos.topicbox.com/groups/developer/members> + delivery options <https://illumos.topicbox.com/groups/developer/subscription>Permalink <https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-Mde1d1df4cc94d80d8b14f740>

[-- Attachment #2: Type: text/html, Size: 12888 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 6134 bytes --]

Thanks Toomas,
 
is the split possible also on rpool?

Also, the split command rename the detached disk with a new pool name, e.g. rpool1, while I thought it would remain unmounted and with the original pool name.
This may create some confusion, as I won't be able to import this and rename it to rpool anywhere an rpool is alive, unless I boot from alternate media, which is exactly the problem I'm facing: no iso for sparc I made recently can boot properly on T4, can't say why, so I want to have copies of production rpool as alternate boot disks, or to boot other machines.
Will the splitted and renamed rpool1 disk boot correctly? Or should I do something to make the BE aware of the new name?
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 7 settembre 2023 9.21.36 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 7. Sep 2023, at 09:54, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
I bet it is the same reason....that "NOTICE: Performing full ZFS device scan!" never happened on x86 in the same situation.
That means I have to remove the disk from the mirror when physically removed or failed.
 


 
ZFS is managing configurations differently (we have 4 copies of labels per disk), so that assumption does not hold there.
 
If you want to re-use half of the mirror in other system, then you would need to not just pull the disk out, but use zpool split and export it before disk removal.
 
Why the “broken mirror” fails to boot on SPARC — the one possible issue may be, the zfs boot block fails to try alternate block address when first one fails. In case of mirror, we have data stored as multiple copies and if the zfs reader will try missing copy first, it may be it fails to get to next one. But that is just an guess about what might be happening there (zfs reader is written in forth in case of SPARC). Of course, this is something we could improve, but since we have announced, we will not support SPARC… 
 
So, for workaround, this situation would mean that if you have broken disk in boot pool, you can use hot replace if hw does support it (zpool offline the disk, cfgadm -c unconfigure disk, replace it, then cfgadm -c configure, zpool replace). Or if hot replace is not possible and boot is failing, then boot from alternate media, import and resilver the pool and that should fix the pool.
 
rgds,
toomas
 

Probelm is, if I detach the disk, that disk won't be anymore usable alone for a second machine, so I will detach it from the mirror after it has been marked as degraded.
That removed disk, though, won't be able to boot alone on another machine, I need some way to transform it into a single disk pool.
Maybe I can import it on another running system, detach the missing disk and export it before booting it on another system?
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 6 settembre 2023 18.17.15 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 6. Sep 2023, at 18:14, Peter Tribble <peter.tribble@gmail.com> wrote:
On Wed, Sep 6, 2023 at 3:46 PM Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
Looks like a very similar problem to this, though here he was using SVM...:

 
I remember that one. That was a plain and simple bug in SVM (it required you to have
a majority of metadevice dbs in order to boot, and exactly half isn't a majority - we had
a process where if a disk failed we simply force deleted the metadevice dbs on the
failed device and then everything's happy).



 
 
it was not bug but documented feature and the reason was about majority election mechanism used to decide if configuration is usable. So to survive, you needed to have 50% of metadb available and to boot, you needed 50% + 1. So with 2 disks setup, you had to split the number of metadb copies between 2 disks, and when losing one disk, before shutdown you had to remove instance(s) from failing disks to be able to boot:)
 
rgds,
toomas

 
I don't know what the problem is here; all I would say that the specific consconfig error
you're seeing doesn't mean anything - I think the system is panicing very early, before
everything else has been properly initialized.
 
https://www.experts-exchange.com/questions/27381310/After-reboot-we-can-t-get-it-to-boot-to-root-mirror-Sun-Fire-280R.html
 
problem seems very similar.
I can't believe anyone ever tried to remove a disk from a mirror and boot back from the only one present, on SPARC, or mayb it's just about the T4...
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 

 


Da: Toomas Soome via illumos-developer <developer@lists.illumos.org>
A: illumos-developer <developer@lists.illumos.org>
Data: 5 settembre 2023 14.17.14 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics





On 5. Sep 2023, at 15:12, Gabriele Bulfon via illumos-developer <developer@lists.illumos.org> wrote:
NOTICE: Finished PCI probing.
ttya not found.
ttya not found.
SPARC T4-1, No Keyboard
 



thats interesting, if you get to ok prompt, does devalias command list ttya?
 
rgds,
toomas



-- 
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/







illumos / illumos-developer / see discussions + participants + delivery options Permalink

[-- Attachment #1.2: Type: text/html, Size: 16259 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Udo Grabowski (IMK)]

[-- Attachment #1: Type: text/plain, Size: 1301 bytes --]

On 18/09/2023 11:26, Gabriele Bulfon via illumos-developer wrote:
> Thanks Toomas,
>
> is the split possible also on rpool?
>
> Also, the split command rename the detached disk with a new pool name, e.g.
> rpool1, while I thought it would remain unmounted and with the original pool name.
> This may create some confusion, as I won't be able to import this and rename it
> to rpool anywhere an rpool is alive, unless I boot from alternate media, which
> is exactly the problem I'm facing: no iso for sparc I made recently can boot
> properly on T4, can't say why, so I want to have copies of production rpool as
> alternate boot disks, or to boot other machines.
> Will the splitted and renamed rpool1 disk boot correctly? Or should I do
> something to make the BE aware of the new name?
>

Yes, the rpool1 will boot, I had this (unnoticed) for a couple of
days when cloning a machine with a mirror split. But renamed it
someday by booting into live media, so that I don't have to change
scripts running in the cluster.
-- 
Dr.Udo Grabowski   Inst.f.Meteorology & Climate Research IMK-ASF-SAT
http://www.imk-asf.kit.edu/english/sat.php
KIT - Karlsruhe Institute of Technology           http://www.kit.edu
Postfach 3640,76021 Karlsruhe,Germany T:(+49)721 608-26026 F:-926026


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 5804 bytes --]

Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

[Gabriele Bulfon]

[-- Attachment #1.1: Type: text/plain, Size: 2027 bytes --]

Thanks! Worked like a charm!
 
Gabriele
 
 
Sonicle S.r.l. : http://www.sonicle.com
Music: http://www.gabrielebulfon.com
eXoplanets : https://gabrielebulfon.bandcamp.com/album/exoplanets
 




----------------------------------------------------------------------------------

Da: Udo Grabowski (IMK) <udo.grabowski@kit.edu>
A: developer@lists.illumos.org
Data: 18 settembre 2023 11.39.43 CEST
Oggetto: Re: [developer] Booting sparc T4 with just one of the mirrored rpool disks panics

On 18/09/2023 11:26, Gabriele Bulfon via illumos-developer wrote:> Thanks Toomas,>> is the split possible also on rpool?>> Also, the split command rename the detached disk with a new pool name, e.g.> rpool1, while I thought it would remain unmounted and with the original pool name.> This may create some confusion, as I won't be able to import this and rename it> to rpool anywhere an rpool is alive, unless I boot from alternate media, which> is exactly the problem I'm facing: no iso for sparc I made recently can boot> properly on T4, can't say why, so I want to have copies of production rpool as> alternate boot disks, or to boot other machines.> Will the splitted and renamed rpool1 disk boot correctly? Or should I do> something to make the BE aware of the new name?>Yes, the rpool1 will boot, I had this (unnoticed) for a couple ofdays when cloning a machine with a mirror split. But renamed itsomeday by booting into live media, so that I don't have to changescripts running in the cluster.-- Dr.Udo Grabowski   Inst.f.Meteorology & Climate Research IMK-ASF-SAThttp://www.imk-asf.kit.edu/english/sat.phpKIT - Karlsruhe Institute of Technology           http://www.kit.eduPostfach 3640,76021 Karlsruhe,Germany T:(+49)721 608-26026 F:-926026------------------------------------------illumos: illumos-developerPermalink: https://illumos.topicbox.com/groups/developer/Td184b2cd65df4901-M5447b4a2313b24d7d02b534cDelivery options: https://illumos.topicbox.com/groups/developer/subscription

<<smime.p7s>>

[-- Attachment #1.2: Type: text/html, Size: 3342 bytes --]