[OpenIndiana-discuss] debugging service startup errors after recent pkg update

[Andreas Wacknitz]

Am 15.06.25 um 11:08 schrieb Mooney, Tim via openindiana-discuss:
> In message <DS0PR08MB914059C23CA558E003A03D8CED76A at DS0PR08MB9140.namprd08.prod.
> outlook.com>, "Mooney, Tim via openindiana-discuss" writes:
>> I'm looking for advice and suggestions on how to potentially rescue the system
>> .  In particular, if I  boot off a recent hipster live USB, is it possible to
>> access the boot environment on my system's disks?  I'm familiar (in theory) wi
>> th the idea of booting into an alternate boot environment and potentially addi
>> I see no harm in booting the latest OI ISO to see if you can spot
>> a hardware failure.
> Slow progress, but a few potentially useful datapoints...
>
> If I boot the workstation off the 2025.04 Live USB, it boots quite quickly and what I've been able to try all works.  The Live USB does display the lightdm GUI at fairly low resolution, but when I try log in with jack/jack it disappears for a couple seconds and then returns me to the login box.
The 2025.04 gui-installer images are known to be broken (jack's home 
folder isn't owned by jack).
Please use newer images, eg. from 
https://dlc.openindiana.org/isos/hipster/20250606/
(or even https://dlc.openindiana.org/isos/hipster/test/).

>
> I'm fairly certain the lightdm issue is because the latest Live USB has nvidia drivers (470) that are too new for the Nvidia Quadro NVS that's in the workstation.  I had to lock the driver/nvidia at the 390 version for the card to continue to be supported.
>
> Later today (after I've had some sleep) I'll try an older Live USB to see if one from 2023 or so ships with the nvidia-390 version.
>
> If I boot the Live USB into single user mode, a 'zpool import' does see the rpool from the problematic install.  I didn't have good documentation for zpool import when I was trying this earlier, so I ended up doing:
>
>      zpool import -f rpool -R /mnt/rpool/
>
> That did succeed and I could see the non-OS datasets for /usr/local , the home dir filesystem and one other, but what I didn't see was /, /usr, or /var.
>
> What was a little surprising was that once I had booted the Live USB in single user mode and imported the workstation's rpool to the /mnt/rpool/ area, 'beadm list' could see all of the boot environments from the workstation, without having to specify any arguments to tell it where the alternate location was for the rpool.
>
> One other thing stuck out about the boot environments.  I'm not great about cleaning up old boot environments so I have a bunch of them.  Most of them are in the 50-250 MB range in size.  The boot environment that got created as part of the 'pkg update' that preceded this mess says its 575 GB (not MB, GB).  That seems suspicious.
>
> Also, while searching for help on Illumos/Solaris boot hangs, I found the Oracle docs for passing "-m milestone=none" via the boot command.   Booting the workstation (not the Live USB) with
>
>      boot -m milestone=none
>
> did allow me to get a system maintenance prompt and get logged in as root.  From there "zpool status" outputs the first few status lines right away, hangs for about 3-4 minutes, and then outputs the config for my rpool.  The two disk devices, the "mirror-0" mirror device, and rpool all say "ONLINE".
>
> I've started a "zpool scrub rpool", and somewhat surprisingly now when I run "zpool status" it only hangs for a few seconds between outputting the pool/state/status/action/scan output and then outputting the config.  Prior to starting the scrub, that pause was much, much longer.
>
> I'm going to let the scrub run while I sleep.
>
> Any thoughts anyone has about how to proceed next would be greatly appreciated.
>
> Thanks,
>
> Tim
> _______________________________________________
> openindiana-discuss mailing list
> openindiana-discuss at openindiana.org
> https://openindiana.org/mailman/listinfo/openindiana-discuss