omniosorg / kayak Goto Github PK

Booting into the installer and escaping to shell, I wanted to see the kernel messages and wasn't unable to -- /var/adm/messages is missing, and /bin/dmesg is linked to /sbin/dmesg, which is simply isn't there.

Tried omniosce-bloody-20190703.iso, and understand that it's unsupported version, though the issue should be common?

Hello,

I'm having installation problems on my setup, but what's really hard right now is I can't seem to find any logs from kayak, and anything helpful flashes up on-screen very briefly. Is there a default log file that kayak uses when booting from the iso?

From here in install_help and here in rpool-install it would seem there should be a file under /tmp/kayak.log - but I can't find this, or any file called kayak.log on the system.

Am I missing something here or are my logs missing!?

Thanks!

I just installed a new r151030 host. While installing it I got to the configuration part where you give it a name and it failed on the domain name, only taking a small part of the domain. Instead of leaving it with a small segment of the domain I was forced to leave the domain entry blank. This left me with a host configured as hostname.local instead of hostname.chscc.biostat.washington.edu.

As I recall it cut off at the n in washington but that was yesterday and I've slept since then so consider that detail unreliable.

I should note this is with a slightly old 151030 install disk. I didn't bother to grab a new one just for one host. I think the installer was 'h' so this may have been reported/fixed since then.

According to this message:

From: Andy Fiddaman [email protected]
To: illumos-discuss
Subject: Re: [discuss] illumos / OmniOS boot loader development

On Fri, 26 May 2023, Jason King wrote:

You currently can?t run illumos as a Hyper-V Gen2 VM. There?s a number of
unique things a Gen2 VM must support that illumos currently does not.

That's true for illumos, but OmniOS runs on Hyper-V Gen2 as it has drivers contributed by Delphix which are not yet upstream.
All new OmniOS releases are boot tested on Hyper-V Gen1 and Gen2 via an installation from .iso.

it should be possible to install omnios-r151046.iso onto a Hyper-V Gen 2 VM. It fails for me on Windows 11 Pro, with a selection of 2048 MB or 4096 MB and with between 1 and 5 Virtual processors.

With verbose selected the last few lines displayed on the console are:

ramdisk0 at root
ramdisk0 is /ramdisk
root on /ramdisk:a fsytpe ufs
/cpus (cpunex0) online
pseudo-device: dld0
dld0 is /pseudo/dld@0

While attempting to install OmniOSCE in a VM from a virtio-block (or nvme) device, it was unable to locate the installation media.

Comments and logic in mount_media.c tell the tale:
https://github.com/omniosorg/kayak/blob/master/src/mount_media.c#L242-L248

/*
 * Callback function for di_walk_minor. For each node that appears to match
 * our criteria (a USB block device, or a CD), mount it and see if it
 * matches the volume set id passed on the command line. If so, we're done
 * and can terminate the walk. In all error cases, just continue walking the
 * tree.
 */

The particular hypervisor I'm installing under (propolis) lacks support for USB or CDROM devices currently, which makes installation via kayak seemingly impossible at the moment.

the BE of a freshly installed OmniOS has no UUID:

hadfl@mars:~$ beadm list -H
omnios;;;;27661312;static;1499278066

Therefore e.g. installing a native zone fails:

hadfl@mars:~$ sudo zoneadm -z testzone install
Sanity Check: Looking for 'entire' incorporation.
ERROR: unable to determine global zone boot environment.
ERROR: cannot install zone testzone

In earlier installers, OmniOS was installed to omniosvar and then a new BE has been created after installation.

Workaround until fixed:
Create a new BE manually after installing OmniOS.

The recently-added lmrc(7D) driver should be added to the miniroot so that drives behind the controllers it supports may be seen during install, etc.

It would be nice if Kayak would work with a root pool that has the same name as a directory that exists in the system, such as system, root or dev.

In this case, it will have to pick a different mount-point for the pool.

some additional packages have to be installed to run kayak:
$ sudo pkg install gnu-make cdrtools gnu-tar

The start method of the kayak service does an "import cherrypy". However, nothing insures that the Python3 version of cherrypy is even installed. If it isn't, svc:/network/kayak:default will never transition to online.

So, please add

depend fmri=pkg:/library/python-3/[email protected] type=require

to the manifest of system/install/kayak.

I believe I've run into a regression going from the r151028 to the r151030 installer, in that the option to force 4K sector sizes for the root pool is no longer honored: after rebooting into the new system, I find that the ashift for the pool is 9 (with r151028, it's correctly being set as 12). I originally noticed this using physical hardware, but I'm able to also reproduce it under VirtualBox. (Also, I'm using omniosce-r151028b.iso for r151028 and omniosce-r151030h.iso for r151030.)

Let me know if there are any additional details I can provide (please bear with me, though, I'm currently a bit of novice when it comes to OmniOS)!

Hi all,

I think it is not very easy for new users coming to Omnios from various other distros to make the initial network configuration so I have written a small tool in Python to remedy this obstacle. The script is tested and runs on all Omnios as of r151014. I have tested on r151014 and r151022

The script is not ment to be the swiss army knife for configuring networks on Omnios so there is no support for vnic, vlan, and aggregates. It will only support configuring a single nic given the
user the option for dhcp and static setup. The user is also asked whether to configure dns in which case the user is asked for an IP.

Currently it will only handle ethernet but IPoIB is on my todo list. The script can be downloaded here: http://git.datanom.net/netconf.git/

Installing OmniOS on NVMe Samsung 960 Pro, having set Force 4K Sectors to YES, OmniOS installer signals:

Cannot unload module: sd
Will be unloaded upon reboot.
Forcing update of sd.conf.
failed to read label from /dev/dsk/c4t0025385971B16535d0s0

Checking post-install, ashift is indeed 9.

Unfortunately I realised this after doing many installs.

The error message is flashing too quickly to recognise and it did not come to me to double check until now.

By the way, many SSD, including Samsung 950, 960, 970 are 8K drives, it seems they should have ashift=13.

It would be nice to have option to force either 4K or 8K.

Thank you.

Hope this is the right place to post to (I'm just having my initial contact with the Illumos world). I downloaded an ISO for r151026 and tried to do an installation in VirtualBox. Out of curiosity I chose the old installer and when I selected "Find disks, create rpool, and install OmniOSce", an error message shows up for the blink of an eye. I kept the "1" key pressed so that the message was displayed over and over again so that I could finally read it:

/kayak/installer/kayak-menu[254]: /kayak/installer/find-and-install: not found [No such file or directory]

It looks like the script is present here in the repo. But obviously there's something wrong on the install media.

I rebooted and used the new installer (which is quite nice, BTW!) to get OmniOS put on the virtual disk. Worked like a charm. Still I thought that I might bring this to the developer's attention in case it hasn't been known so far (only newbies don't do a custom installation, right? ;)).

SYSTEM in build/build_miniroot lists system/boot/wanboot and system/boot/wanboot/internal which are obsolete for a long time.

This is going to be in run-on shorthand because I originally wrote the following block of text as a title.

Newly installed OmniOS bloody (claims to be 151031) system, had set root and user account passwords in installer, attempting to log in either at console or at SSH threw an error about empty password (not disabled login, or no password which I assume would just let me log in without a password on /dev/console, but a PAM error), ended up rebooting into the installer, opening up a shell, mounting and chrooting to the system, and since passwd returned EPERM, performing surgery using /bin/ed on /etc/shadow to add a password hash (derived from the output of an IRC server's MKPASSWD extension, funnily enough) to the user account I created. This allowed me to ssh into the machine after rebooting into the new install. sudo passwd root (I had given the user account sudo permissions) allowed me then to set root's password as if nothing had ever happened.

USB install on a Simply NUC Topaz 2 i7. System was booted from USB with no user interaction on the keyboard and hangs with the following displayed:

`OmniOS r151046 Version omnios-r151046-0c94315846 64-bit
Copyright (c) 2012-2017 OmniTi Computer Consulting, Inc.
Copyright (c) 2017-2023 OmniOS Community Edition (OmniOSce) Association.
WARNING: xhci0: encountered port capabilities with unknown USB version: 3.20

Configuring devices.
Applying initial boot settings...

Scanning for media...

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!! Could not mount installation media !!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

`
This may be caused by an underlying driver issue. omnios-151044k.usb.dd also fails to install

Can the OmniOSce installer also present an option to enable support for virtual terminals?
The following (by Volker Brandt) can be used.

svcs console-login
svcadm enable vtdaemon
svcadm enable console-login:vt2
svcadm enable console-login:vt3
svcadm enable console-login:vt4
svcadm enable console-login:vt5
svcadm enable console-login:vt6
svcs console-login vtdaemon
svccfg -s vtdaemon setprop options/hotkeys=true
svccfg -s vtdaemon setprop options/secure=false
svcadm refresh vtdaemon
svcadm restart vtdaemon

PXE Boot fails on a Simply NUC Topaz 2 i7. Displays NBP file downloaded successfully. then drops through to the next boot option, without trying to get any further files using tftp. Both pxeboot and pxegrub fail. Attached is a Wireshark capture for pxeboot (the capture for pxegrub is similar except for the packet count so is not attached). The server (192.168.88.1) and topaz (192.168.88.2) are connected back-to-back on a single cable.

Possibly the same underlying problem as omnios-r151046-rc3.usb-dd fails on Simply NUC Topaz 2 i7.

I have successfully PXE booted Ubuntu Server 23.04 (Lunar Lobster) as far as the selection screen on the same configuration.

topaz.pcapng.gz

Dunno if this is the right tracker, but https://illumos.org/docs/user-guide/bugs/ pointed me here and I didn't find anything else, so

I used https://downloads.omnios.org/media/stable/omniosce-r151036m.iso, installed, pressed the configure your new omnios installation button, then toggled extra packages. It breaks more going one way than the other, but appears to function either way, confer screenshot: