Trouble installing openSUSE-Tumbleweed

[George R Goffe]

Per,

VirtualBox can be downloaded as an rpm from http://download.virtualbox.org/virtualbox. Scroll down to the 6.1.4 directory. There are two Suse related rpms:

      VirtualBox-6.1-6.1.4_136177_openSUSE132-1.x86_64.rpm                   19-Feb-2020 20:18  96M
      VirtualBox-6.1-6.1.4_136177_openSUSE150-1.x86_64.rpm                   19-Feb-2020 20:18  89M

It's easy to install... just "rpm -ivh". A post.install scriptlett tries to assemble kernel modules for your "current" kernel.

Just run "/usr/bin or /bin/virtualbox. You'll have to config a VM with the GUI that pops up... Config memory (2g or 4g), virtual disk size, network (bridged), storage (optical device pointing to a .iso image). It's fairly simple to config... Then poke the green arrow (start). You get a virtual console and away you go.

Back to the original "problem".

If I understand VB correctly, the memory used by the VM is controlled by what you configure when you define the VM... Hmmm... thinking... I think you're right... points to a VB problem. The fact that the host OS senses out of memory  for the VM is suspicious. I wonder if VB could detect the VM out of memory and reflect that to the host? I have a running question with the VB user/dev group. I'll ask that question.


I do have a log from VB about the guest. It looks like things were running ok and then there is a gap and then some messages and inexplicable messages and then what looks like a "I'm giving up" message and then termination.

Here's a sample of the msgs. Can I send you a log? It almost looks like the hdd failed or stopped responding.

1219 01:08:56.022924 VMMDev: Guest Log: int13_harddisk: function 02, unmapped device for ELDL=8f
1220 01:08:56.227979 GIM: KVM: VCPU  0: Enabled system-time struct. at 0x0000000076401000 - u32TscScale=0xe942f1f9 i8TscShift=-1 uVersion=2 fFlags=0x1 uTsc=0x48fe84cfd uVirtNanoTS=0x3c2881220 c98f69
1221 01:08:56.228030 TM: Switching TSC mode from 'Dynamic' to 'RealTscOffset'
1222 01:08:56.526855 TMR3UtcNow: nsNow=1 584 727 161 285 063 000 nsPrev=1 584 723 026 890 398 000 -> cNsDelta=4 134 394 665 000 (offLag=0 offVirtualSync=0 offVirtualSyncGivenUp=0, NowAgai1222 n=1 584 727 161 285 063 000)
1223 01:08:56.526886 GIM: KVM: Enabled wall-clock struct. at 0x0000000076400000 - u32Sec=1584727161 u32Nano=285063000 uVersion=2
1224 01:08:56.569601 PIT: mode=2 count=0x12a5 (4773) - 249.98 Hz (ch=0)
1225 01:08:56.570178 APIC0: Switched mode to x2APIC
1226 01:08:56.575512 IEM: rdmsr(0x140) -> #GP(0)
1227 01:08:56.700730 PIT: mode=0 count=0x10000 (65536) - 18.20 Hz (ch=0)
1228 01:08:56.901791 OHCI: Software reset
1229 01:08:58.868741 AHCI#0: Reset the HBA
1230 01:08:58.868768 VD#0: Cancelling all active requests
1231 01:08:58.872171 AHCI#0: Port 0 reset
1232 01:08:58.875019 VD#0: Cancelling all active requests
1233 01:09:00.511613 PIIX3 ATA: Ctl#1: RESET, DevSel=0 AIOIf=0 CmdIf0=0xa0 (-1 usec ago) CmdIf1=0x00 (-1 usec ago)
1234 01:09:00.511689 PIIX3 ATA: Ctl#1: finished processing RESET
1235 01:09:00.897101 OHCI: USB Reset
1236 01:09:00.961659 OHCI: Software reset
1237 01:09:00.961959 OHCI: USB Operational
1238 01:09:01.602155 Enabling different vbva mode
1239 01:09:01.861965 Display::i_handleDisplayResize: uScreenId=0 pvVRAM=00007f8d84000000 w=800 h=600 bpp=32 cbLine=0xC80 flags=0x1 origin=0,0
1240 03:39:24.558347 AsyncCompletion: Task 0x007f8e182ba9c0 completed after 11 seconds

Is there any throttling done at the suse distribution site? Like amount of download?

I can send you the log just in case anything there would give you an idea?

George...

Per,
The partitioning is one major area. I select a lot of the products...
multiple desktop managers... rename the ethernet device name (eth0)...
I'm looking at the partially(?) installed system via the rescue option
from the .iso. I don't see any way to see where the installation was
halted. It all looks ok so far, /boot is almost empty.

Hi George

the latter is not good - a a brandnew installation, you ought to have at
one initrd and one kernel, symlinked as initrd and vmlinuz.  So clearly
it did not finish.

Do you have a way to tell where the install failed?

Maybe check out the logs /var/log/YaST2/y2log  (something like that).

It's basically not so interesting when the VM is being killed - to me,
that puts the blame entirely on the VM setup.

As it is now, the install appears to be running properly... the
packages fly on by on the GUI. I step away for a few hours and come
back to find the installation aborted. /var/log/messages shows the
oom-killer has been invoked against the VM instance.

I'm not familiar with VirtualBox, but does it not have a specific amount
of memory, with a hard upper limit?  Well, apparently not, so maybe add
some memory (provided your host has enough).  I'm certain an
installation of Leap takes no more than 2Gb or even 1Gb with active
swap.
--
Per Jessen, Zürich (11.1°C)
http://www.hostsuisse.com/ - virtual servers, made in Switzerland.


--
To unsubscribe, e-mail: opensuse+***@opensuse.org
To contact the owner, e-mail: opensuse+***@opensuse.org

[George R Goffe]

Per,

Here's what I sent to David who also responded to my plea for help.

I do NOT think it's the VB config. I've been able to install/run MANY versions of Fedora Core. I suspect that it's my config options that are breaking the install.

Is there a way to dump the generated "proposal" to a file?

David,

Thanks for responding.

I have used a LOT of the options for the install but this is NOT new for me.

I just tried a "production" version of VB and it failed as well. Same result.

I'm looking at /var/log/messages. It seems that oom-reaper was invoked due to VB. I'm beginning to think that something in my config is breaking the install or rather causing a break.

Is there a way to get all the config parameters dumped to a file? Like I said, my config is complicated.

Maybe I'll try Tumbleweed 13x.

Best regards,

George...

Hi,
I've been trying to install several versions of Tumbleweed on my
system here and each has failed. Miserably.
I'm using the latest developer version of VirtualBox,
VirtualBox-6.1.97-136454-Linux_amd64.run.
I have a log file from the VB Manager. The VM appears to have
vaporized. I'm getting NO indications except for "Aborted" in the VM
entry in the VB Manager.
I even tried Leap with the same results.

There can only be one conclusion - something is wrong with your
VirtualBox setup. 

I have started this install effort with a VM with 2GB Memory defined.
This last attempt I upped the memory to 4GB. ALL with the same
results.

1Gb would be too little, and would require swap, but 2Gb is fine.
--
Per Jessen, Zürich (10.9°C)
http://www.cloudsuisse.com/ - your owncloud, hosted in Switzerland.


--
To unsubscribe, e-mail: opensuse+***@opensuse.org
To contact the owner, e-mail: opensuse+***@opensuse.org

[George R Goffe]

Per,

The partitioning is one major area. I select a lot of the products... multiple desktop managers... rename the ethernet device name (eth0)...

I'm looking at the partially(?) installed system via the rescue option from the .iso. I don't see any way to see where the installation was halted. It all looks ok so far, /boot is almost empty. Do you have a way to tell where the install failed?

As it is now, the install appears to be running properly... the packages fly on by on the GUI. I step away for a few hours and come back to find the installation aborted. /var/log/messages shows the oom-killer has been invoked against the VM instance. This seems to tell me that the installer has a memory problem. "buying" memory but never freeing it. Eventually, the VM is killed.

Again, THANKS for responding.

George...

Post by George R Goffe
Per,
Here's what I sent to David who also responded to my plea for help.
I do NOT think it's the VB config. I've been able to install/run MANY
versions of Fedora Core. I suspect that it's my config options that
are breaking the install.

I run TW and Leap both on xen and on real iron, many of them - sometimes
a config (nfsroot, iscsi, multipathing) can be a little tricky to get
to work, but in the end it works.

Which config options are you referring to here ? 
--
Per Jessen, Zürich (10.7°C)
http://www.hostsuisse.com/ - virtual servers, made in Switzerland.


--
To unsubscribe, e-mail: opensuse+***@opensuse.org
To contact the owner, e-mail: opensuse+***@opensuse.org

[Per Jessen]

Per,
The partitioning is one major area. I select a lot of the products...
multiple desktop managers... rename the ethernet device name (eth0)...
I'm looking at the partially(?) installed system via the rescue option
from the .iso. I don't see any way to see where the installation was
halted. It all looks ok so far, /boot is almost empty.

Hi George

the latter is not good - a a brandnew installation, you ought to have at
one initrd and one kernel, symlinked as initrd and vmlinuz. So clearly
it did not finish.

Do you have a way to tell where the install failed?

Maybe check out the logs /var/log/YaST2/y2log (something like that).

It's basically not so interesting when the VM is being killed - to me,
that puts the blame entirely on the VM setup.

As it is now, the install appears to be running properly... the
packages fly on by on the GUI. I step away for a few hours and come
back to find the installation aborted. /var/log/messages shows the
oom-killer has been invoked against the VM instance.

I'm not familiar with VirtualBox, but does it not have a specific amount
of memory, with a hard upper limit? Well, apparently not, so maybe add
some memory (provided your host has enough). I'm certain an
installation of Leap takes no more than 2Gb or even 1Gb with active
swap.

[George R Goffe]

Carlos,

The debug output came out because I added '-xv' at the beginning of the script.

Well... some good news and bad news.

The good: I used a newer iso image and this problem seems to have gone away.

The bad: The installer has a program blkid which is invoked when installing the kernel-default. This installation process runs dracut which runs blkid. blkid reveals (I believe) a kernel bug that appears when it (blkid) accesses a swap area... it goes into an uninterruptible sleep... which is the END OF THE INSTALL. From the bug below you'll see that blkid w/o operands hangs and shutdown now (init 0) also hangs when it tries to turn off swap. Sigh...

There's a bug report about this that I wrote at https://bugzilla.redhat.com/show_bug.cgi?id=1800831. Somehow I didn't find out what the root cause was... I just wrote the report and provided info that was requested. I think the problem was in the kernel though.

I'm running this kernel now because VirtualBox can build it's LKMs on it (5.5.0-0.rc6.git0.1.fc32.x86_64, the rest of Fedora Core 33 (Rawhide) is at 5.6.0-0.rc6.git2.1.fc33), the Tumbleweed installer is at 5.5.9-1.1 and still(?) has the bug.

If I can help further please let me know.

Best regards,

George...




On Saturday, March 21, 2020, 5:31:16 PM PDT, Carlos E. R. <***@telefonica.net> wrote:





-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I had to copy the file
https://build.opensuse.org/source/openSUSE:Factory/susepaste/susepaste-script-0.6.tar.bz2?rev=51f7f0831a45466f15e3e833efa95d69
and manually install it.

I was rather thinking of:
<http://download.opensuse.org/tumbleweed/repo/oss/noarch/susepaste-0.6-2.1.noarch.rpm>

but nevermind.

susepaste -t "tumbleweed install oom - host msgs" -n grgoffe -e 40320 -f text  tumbleweed-oom-latest-abort.messages

The "-f text" is not needed, it is supposed to detect the file type.
Nevermind. What I don't understand is why you get the verbose or debug
output.

+ expr $'https://susepaste.org/95061758\r' : '^https://susepaste.org/[0-9a-f]\+'
++ echo $'https://susepaste.org/95061758\r'
++ sed 's|^https://susepaste.org/\([0-9a-f]\+\)[^0-9a-f]*|\1|'
+ ID=95061758
+ echo 'Pasted as:'
+ echo '   https://susepaste.org/95061758'
https://susepaste.org/95061758
+ echo '   https://paste.opensuse.org/95061758'
https://paste.opensuse.org/95061758
+ '[' -x /usr/bin/xclip ']'
+ echo https://susepaste.org/95061758
+ xclip -selection XA_CLIPBOARD
+ echo 'Link is also in your clipboard.'
Link is also in your clipboard.
When I go to either of the URLs above I get page not found. I try to
Authentication error; not a valid OpenID.

I don't see it here, either:
<https://susepaste.org/lists>

I don't see how this script allows for authentication with the site.
Anonymous posting? Sounds dangerous to DOS attacks?

There is no login, all is anonimous. It is the same as publishing a post
here, it is public.

Well, there are other pastebin sites, if you want to try.

I must be doing something wrong but do NOT see it.

No, I do not know why it failed, sorry.

- --
Cheers
        Carlos E. R.

        (from openSUSE 15.1 (Legolas))

-----BEGIN PGP SIGNATURE-----

iHoEARECADoWIQQZEb51mJKK1KpcU/W1MxgcbY1H1QUCXnaxnhwccm9iaW4ubGlz
dGFzQHRlbGVmb25pY2EubmV0AAoJELUzGBxtjUfVZHwAoIV1kjJZeSt5julIpqKa
C6QtL878AJ9X5WMmcXqVi9wkNIqfRRFdhOPo5A==
=mu9N

-----END PGP SIGNATURE-----

[Per Jessen]

Post by George R Goffe
The bad: The installer has a program blkid which is invoked when
installing the kernel-default. This installation process runs dracut
which runs blkid. blkid reveals (I believe) a kernel bug that appears
when it (blkid) accesses a swap area... it goes into an
uninterruptible sleep... which is the END OF THE INSTALL. From the
bug below you'll see that blkid w/o operands hangs and shutdown now
(init 0) also hangs when it tries to turn off swap. Sigh...

If this is a bug in blkid that appears during the building of the
initrd, it ought to be reproducible on virtually every install ?
Maybe I'll crank up a xen guest with TW.

I forgot, only just yesterday I upgraded a TW desktop, new kernel 5.5.9,
so initrd was rebuilt. No problem.

Running 'blkid' without arguments:

# blkid
/dev/sda1: UUID="ca3b748f-7ce5-4e83-8a35-7801a6a7d898" TYPE="swap"
PARTUUID="837f9747-01"
/dev/sda2: UUID="3f44a97f-c1cf-4e45-b66e-959881f194fe" TYPE="jfs"
PARTUUID="837f9747-02"

When your 'blkid' gets hung up, it must be environment dependent. My
new xen guest with TW is just about ready, I'll report later.

[Per Jessen]

Post by Per Jessen
When your 'blkid' gets hung up, it must be environment dependent. My
new xen guest with TW is just about ready, I'll report later.

On new TW guest:

# blkid
/dev/xvda1: UUID="d50c07b1-bbb4-4eae-b90d-5ae1cc8e1836" TYPE="jfs"
PARTUUID="00079a3b-01"
/dev/xvda2: UUID="36dab3aa-8baf-4284-80b3-9b56fd29ed88" TYPE="swap"
PARTUUID="00079a3b-02"

[Carlos E. R.]

Carlos,
The debug output came out because I added '-xv' at the beginning of the script.

Ah! Ok :-)

Well... some good news and bad news.
The good: I used a newer iso image and this problem seems to have gone away.
The bad: The installer has a program blkid which is invoked when
installing the kernel-default. This installation process runs dracut
which runs blkid. blkid reveals (I believe) a kernel bug that appears
when it (blkid) accesses a swap area... it goes into an
uninterruptible sleep... which is the END OF THE INSTALL. From the
bug below you'll see that blkid w/o operands hangs and shutdown now
(init 0) also hangs when it tries to turn off swap. Sigh...

Curious. And this was not caught by openqa :-?

There's a bug report about this that I wrote at
https://bugzilla.redhat.com/show_bug.cgi?id=1800831. Somehow I didn't
find out what the root cause was... I just wrote the report and
provided info that was requested. I think the problem was in the
kernel though.

Well, there needs to be a bug declared at bugzilla.opensuse.org. I'd
guess that if redhat has solved the problem they have not told upstream
and has not propagated.

If you are willing:

<https://en.opensuse.org/openSUSE:Submitting_bug_reports>


I'm in the process of rebuilding/migrating my desktop computer, then I
can try to install TW in VBox and try reproduce your problem. The
existing machine did not have enough RAM.

I'm running this kernel now because VirtualBox can build it's LKMs on
it (5.5.0-0.rc6.git0.1.fc32.x86_64, the rest of Fedora Core 33
(Rawhide) is at 5.6.0-0.rc6.git2.1.fc33), the Tumbleweed installer is
at 5.5.9-1.1 and still(?) has the bug.
If I can help further please let me know.
Best regards,
George...

Er... just one detail, if you allow me :-)

The custom in this list is to reply interlined, as I do in these posts,
not top-posting ;-)

[Carlos E. R.]

Per,
The partitioning is one major area. I select a lot of the products...
multiple desktop managers... rename the ethernet device name (eth0)...
I'm looking at the partially(?) installed system via the rescue option
from the .iso. I don't see any way to see where the installation was
halted. It all looks ok so far, /boot is almost empty. Do you have a way
to tell where the install failed?
As it is now, the install appears to be running properly... the packages
fly on by on the GUI. I step away for a few hours and come back to find
the installation aborted. /var/log/messages shows the oom-killer has
been invoked against the VM instance.

On the host? Then your host system has a problem, because if the
installation system "inside" has a memory problem it is the virtual
machine "inside" which has to call OOM and kill itself - but that would
never affect the virtualization system. If it is VB in the host which
crashes, you have a memory issue on your host, or VB is faulty.

This seems to tell me that the
installer has a memory problem. "buying" memory but never freeing it.
Eventually, the VM is killed.

Ideas.

You could partition the virtual hard disk in advance, creating a swap
partition. Early on the installation, you have to open a terminal (ctrl-X,
I think, or jump to one of the virtual terminals), and activate that swap
partition. That way you can have more memory inside the installation
system.

Then on that same terminal run "top" sorted by memory to watch if
something is growing in memory usage.

- --
Cheers
Carlos E. R.

(from openSUSE 15.1 (Legolas))

[Carlos E. R.]

I had to copy the file
https://build.opensuse.org/source/openSUSE:Factory/susepaste/susepaste-script-0.6.tar.bz2?rev=51f7f0831a45466f15e3e833efa95d69
and manually install it.

I was rather thinking of:
<http://download.opensuse.org/tumbleweed/repo/oss/noarch/susepaste-0.6-2.1.noarch.rpm>

but nevermind.

susepaste -t "tumbleweed install oom - host msgs" -n grgoffe -e 40320 -f text  tumbleweed-oom-latest-abort.messages

The "-f text" is not needed, it is supposed to detect the file type.
Nevermind. What I don't understand is why you get the verbose or debug
output.

+ expr $'https://susepaste.org/95061758\r' : '^https://susepaste.org/[0-9a-f]\+'
++ echo $'https://susepaste.org/95061758\r'
++ sed 's|^https://susepaste.org/\([0-9a-f]\+\)[^0-9a-f]*|\1|'
+ ID=95061758
+ echo 'Pasted as:'
+ echo '   https://susepaste.org/95061758'
https://susepaste.org/95061758
+ echo '   https://paste.opensuse.org/95061758'
https://paste.opensuse.org/95061758
+ '[' -x /usr/bin/xclip ']'
+ echo https://susepaste.org/95061758
+ xclip -selection XA_CLIPBOARD
+ echo 'Link is also in your clipboard.'
Link is also in your clipboard.
When I go to either of the URLs above I get page not found. I try to
Authentication error; not a valid OpenID.

I don't see it here, either:
<https://susepaste.org/lists>

I don't see how this script allows for authentication with the site.
Anonymous posting? Sounds dangerous to DOS attacks?

There is no login, all is anonimous. It is the same as publishing a post
here, it is public.

Well, there are other pastebin sites, if you want to try.

I must be doing something wrong but do NOT see it.

No, I do not know why it failed, sorry.

- --
Cheers
Carlos E. R.

(from openSUSE 15.1 (Legolas))

[Carlos E. R.]

Carlos,
I do have the target virtual hdd pre-allocated. I could configure the
install to leave that virtual partition alone and then do a swapadd
during the install. Do you think that would work?

If it is short on memory, yes, it should. The installation minimal memory
is larger than the running system, which may be as low as half a gig.

But I suspect your problem is with the host.

I like this idea about running top, hopefully it's in the installer.
Am I allowed to send log files to this list? I have a log from the host
during the oom and also from VB.

You can add text attachments, there is a size limit. But it is frowned
uppon if bigish. Better to upload to susepaste.org (say for a week or a
month) and post here the link.

There is also a script to do it from Linux (susepaste). It may claims
failure, but then looking at "recent" at the site, you can find your
upload and paste the link.

- --
Cheers
Carlos E. R.

(from openSUSE 15.1 (Legolas))

[George R Goffe]

Carlos,

I had to copy the file https://build.opensuse.org/source/openSUSE:Factory/susepaste/susepaste-script-0.6.tar.bz2?rev=51f7f0831a45466f15e3e833efa95d69 and manually install it.

susepaste -t "tumbleweed install oom - host msgs" -n grgoffe -e 40320 -f text  tumbleweed-oom-latest-abort.messages

produces:

+ expr $'https://susepaste.org/95061758\r' : '^https://susepaste.org/[0-9a-f]\+'
++ echo $'https://susepaste.org/95061758\r'
++ sed 's|^https://susepaste.org/\([0-9a-f]\+\)[^0-9a-f]*|\1|'
+ ID=95061758
+ echo 'Pasted as:'
Pasted as:
+ echo '   https://susepaste.org/95061758'
https://susepaste.org/95061758
+ echo '   https://paste.opensuse.org/95061758'
https://paste.opensuse.org/95061758
+ '[' -x /usr/bin/xclip ']'
+ echo https://susepaste.org/95061758
+ xclip -selection XA_CLIPBOARD
+ echo 'Link is also in your clipboard.'
Link is also in your clipboard.

When I go to either of the URLs above I get page not found. I try to login and enter my userid(?) grgoffe but get this message: Authentication error; not a valid OpenID.

I don't see how this script allows for authentication with the site. Anonymous posting? Sounds dangerous to DOS attacks?

I must be doing something wrong but do NOT see it.

Your thoughts please?

George...






On Saturday, March 21, 2020, 1:29:01 PM PDT, Carlos E. R. <***@telefonica.net> wrote:





-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Carlos,
I do have the target virtual hdd pre-allocated. I could configure the
install to leave that virtual partition alone and then do a swapadd
during the install. Do you think that would work?

If it is short on memory, yes, it should. The installation minimal memory
is larger than the running system, which may be as low as half a gig.

But I suspect your problem is with the host.

I like this idea about running top, hopefully it's in the installer.
Am I allowed to send log files to this list? I have a log from the host
during the oom and also from VB.

You can add text attachments, there is a size limit. But it is frowned
uppon if bigish. Better to upload to susepaste.org (say for a week or a
month) and post here the link.

There is also a script to do it from Linux (susepaste). It may claims
failure, but then looking at "recent" at the site, you can find your
upload and paste the link.

- --
Cheers
        Carlos E. R.

        (from openSUSE 15.1 (Legolas))

-----BEGIN PGP SIGNATURE-----

iHoEARECADoWIQQZEb51mJKK1KpcU/W1MxgcbY1H1QUCXnZ3MBwccm9iaW4ubGlz
dGFzQHRlbGVmb25pY2EubmV0AAoJELUzGBxtjUfVoJoAn0wiqnzDn8Hh/pA8t9kK
sjkaCLUgAJoDnmW8DxVNbsxvis719EwDOn+/9Q==
=hNmQ

-----END PGP SIGNATURE-----

[George R Goffe]

Carlos,

I do have the target virtual hdd pre-allocated. I could configure the install to leave that virtual partition alone and then do a swapadd during the install. Do you think  that would work?

I like this idea about running top, hopefully it's in the installer.

Am I allowed to send log files to this list? I have a log from the host during the oom and also from VB.

Thanks for your response,

George...


On Saturday, March 21, 2020, 6:16:59 AM PDT, Carlos E. R. <***@telefonica.net> wrote:





-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Per,
The partitioning is one major area. I select a lot of the products...
multiple desktop managers... rename the ethernet device name (eth0)...
I'm looking at the partially(?) installed system via the rescue option
from the .iso. I don't see any way to see where the installation was
halted. It all looks ok so far, /boot is almost empty. Do you have a way
to tell where the install failed?
As it is now, the install appears to be running properly... the packages
fly on by on the GUI. I step away for a few hours and come back to find
the installation aborted. /var/log/messages shows the oom-killer has
been invoked against the VM instance.

On the host? Then your host system has a problem, because if the
installation system "inside" has a memory problem it is the virtual
machine "inside" which has to call OOM and kill itself - but that would
never affect the virtualization system. If it is VB in the host which
crashes, you have a memory issue on your host, or VB is faulty.

This seems to tell me that the
installer has a memory problem. "buying" memory but never freeing it.
Eventually, the VM is killed.

Ideas.

You could partition the virtual hard disk in advance, creating a swap
partition. Early on the installation, you have to open a terminal (ctrl-X,
I think, or jump to one of the virtual terminals), and activate that swap
partition. That way you can have more memory inside the installation

system.


Then on that same terminal run "top" sorted by memory to watch if
something is growing in memory usage.

- --
Cheers
        Carlos E. R.

        (from openSUSE 15.1 (Legolas))

-----BEGIN PGP SIGNATURE-----

iHoEARECADoWIQQZEb51mJKK1KpcU/W1MxgcbY1H1QUCXnYTmxwccm9iaW4ubGlz
dGFzQHRlbGVmb25pY2EubmV0AAoJELUzGBxtjUfV8XUAnjQDvecmw0zuxSWrhkiw
jz0pJ3UVAJ9YuhyjpknCovo7yqu15akJe+8Olw==
=Noau
-----END PGP SIGNATURE-----

[Per Jessen]

Post by George R Goffe
Per,
Here's what I sent to David who also responded to my plea for help.
I do NOT think it's the VB config. I've been able to install/run MANY
versions of Fedora Core. I suspect that it's my config options that
are breaking the install.

I run TW and Leap both on xen and on real iron, many of them - sometimes
a config (nfsroot, iscsi, multipathing) can be a little tricky to get
to work, but in the end it works.

Which config options are you referring to here ?

[Carlos E. R.]

Carlos,
I have two files... around 2-3k each. Paste of the smaller has frozen my browser. Sigh...

:-o

Did I miss something? There's no way to directly upload a file?

Yes, the command susepaste. You can download the rpm. It is just a script
and perhaps a man page.

- --
Cheers
Carlos E. R.

(from openSUSE 15.1 (Legolas))

[Per Jessen]

Hi,
I've been trying to install several versions of Tumbleweed on my
system here and each has failed. Miserably.
I'm using the latest developer version of VirtualBox,
VirtualBox-6.1.97-136454-Linux_amd64.run.
I have a log file from the VB Manager. The VM appears to have
vaporized. I'm getting NO indications except for "Aborted" in the VM
entry in the VB Manager.
I even tried Leap with the same results.

There can only be one conclusion - something is wrong with your
VirtualBox setup.

I have started this install effort with a VM with 2GB Memory defined.
This last attempt I upped the memory to 4GB. ALL with the same
results.

1Gb would be too little, and would require swap, but 2Gb is fine.

[George R Goffe]

Carlos,

I have two files... around 2-3k each. Paste of the smaller has frozen my browser. Sigh...

Did I miss something? There's no way to directly upload a file?

Best regards,

George...







On Saturday, March 21, 2020, 1:02:31 PM PDT, Carlos E. R. <***@telefonica.net> wrote:





-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by George R Goffe
Here's a sample of the msgs. Can I send you a log? It almost looks like the hdd failed or stopped responding.

You can upload logs to susepaste.org for some period of time, and post the
link here.

Post by George R Goffe
1219 01:08:56.022924 VMMDev: Guest Log: int13_harddisk: function 02, unmapped device for ELDL=8f
1220 01:08:56.227979 GIM: KVM: VCPU  0: Enabled system-time struct. at 0x0000000076401000 - u32TscScale=0xe942f1f9 i8TscShift=-1 uVersion=2 fFlags=0x1 uTsc=0x48fe84cfd uVirtNanoTS=0x3c2881220 c98f69
1221 01:08:56.228030 TM: Switching TSC mode from 'Dynamic' to 'RealTscOffset'
1222 01:08:56.526855 TMR3UtcNow: nsNow=1 584 727 161 285 063 000 nsPrev=1 584 723 026 890 398 000 -> cNsDelta=4 134 394 665 000 (offLag=0 offVirtualSync=0 offVirtualSyncGivenUp=0, NowAgai1222 n=1 584 727 161 285 063 000)
1223 01:08:56.526886 GIM: KVM: Enabled wall-clock struct. at 0x0000000076400000 - u32Sec=1584727161 u32Nano=285063000 uVersion=2
1224 01:08:56.569601 PIT: mode=2 count=0x12a5 (4773) - 249.98 Hz (ch=0)
1225 01:08:56.570178 APIC0: Switched mode to x2APIC
1226 01:08:56.575512 IEM: rdmsr(0x140) -> #GP(0)
1227 01:08:56.700730 PIT: mode=0 count=0x10000 (65536) - 18.20 Hz (ch=0)
1228 01:08:56.901791 OHCI: Software reset
1229 01:08:58.868741 AHCI#0: Reset the HBA
1230 01:08:58.868768 VD#0: Cancelling all active requests
1231 01:08:58.872171 AHCI#0: Port 0 reset
1232 01:08:58.875019 VD#0: Cancelling all active requests
1233 01:09:00.511613 PIIX3 ATA: Ctl#1: RESET, DevSel=0 AIOIf=0 CmdIf0=0xa0 (-1 usec ago) CmdIf1=0x00 (-1 usec ago)
1234 01:09:00.511689 PIIX3 ATA: Ctl#1: finished processing RESET
1235 01:09:00.897101 OHCI: USB Reset
1236 01:09:00.961659 OHCI: Software reset
1237 01:09:00.961959 OHCI: USB Operational
1238 01:09:01.602155 Enabling different vbva mode
1239 01:09:01.861965 Display::i_handleDisplayResize: uScreenId=0 pvVRAM=00007f8d84000000 w=800 h=600 bpp=32 cbLine=0xC80 flags=0x1 origin=0,0
1240 03:39:24.558347 AsyncCompletion: Task 0x007f8e182ba9c0 completed after 11 seconds

It does seem some hard disk trouble. I would investigate the host logs for
that. Run the long smart test.

- --
Cheers
        Carlos E. R.

        (from openSUSE 15.1 (Legolas))

-----BEGIN PGP SIGNATURE-----

iHoEARECADoWIQQZEb51mJKK1KpcU/W1MxgcbY1H1QUCXnZyqRwccm9iaW4ubGlz
dGFzQHRlbGVmb25pY2EubmV0AAoJELUzGBxtjUfVKw8An3ihFVAtGB4Il71in+y2
t/OaCfAwAKCLIEkVGIZCEY7bCwZXv6PrREKx+A==
=8XqF
-----END PGP SIGNATURE-----

[Carlos E. R.]

Post by George R Goffe
Here's a sample of the msgs. Can I send you a log? It almost looks like the hdd failed or stopped responding.

You can upload logs to susepaste.org for some period of time, and post the
link here.

Post by George R Goffe
1219 01:08:56.022924 VMMDev: Guest Log: int13_harddisk: function 02, unmapped device for ELDL=8f
1220 01:08:56.227979 GIM: KVM: VCPU  0: Enabled system-time struct. at 0x0000000076401000 - u32TscScale=0xe942f1f9 i8TscShift=-1 uVersion=2 fFlags=0x1 uTsc=0x48fe84cfd uVirtNanoTS=0x3c2881220 c98f69
1221 01:08:56.228030 TM: Switching TSC mode from 'Dynamic' to 'RealTscOffset'
1222 01:08:56.526855 TMR3UtcNow: nsNow=1 584 727 161 285 063 000 nsPrev=1 584 723 026 890 398 000 -> cNsDelta=4 134 394 665 000 (offLag=0 offVirtualSync=0 offVirtualSyncGivenUp=0, NowAgai1222 n=1 584 727 161 285 063 000)
1223 01:08:56.526886 GIM: KVM: Enabled wall-clock struct. at 0x0000000076400000 - u32Sec=1584727161 u32Nano=285063000 uVersion=2
1224 01:08:56.569601 PIT: mode=2 count=0x12a5 (4773) - 249.98 Hz (ch=0)
1225 01:08:56.570178 APIC0: Switched mode to x2APIC
1226 01:08:56.575512 IEM: rdmsr(0x140) -> #GP(0)
1227 01:08:56.700730 PIT: mode=0 count=0x10000 (65536) - 18.20 Hz (ch=0)
1228 01:08:56.901791 OHCI: Software reset
1229 01:08:58.868741 AHCI#0: Reset the HBA
1230 01:08:58.868768 VD#0: Cancelling all active requests
1231 01:08:58.872171 AHCI#0: Port 0 reset
1232 01:08:58.875019 VD#0: Cancelling all active requests
1233 01:09:00.511613 PIIX3 ATA: Ctl#1: RESET, DevSel=0 AIOIf=0 CmdIf0=0xa0 (-1 usec ago) CmdIf1=0x00 (-1 usec ago)
1234 01:09:00.511689 PIIX3 ATA: Ctl#1: finished processing RESET
1235 01:09:00.897101 OHCI: USB Reset
1236 01:09:00.961659 OHCI: Software reset
1237 01:09:00.961959 OHCI: USB Operational
1238 01:09:01.602155 Enabling different vbva mode
1239 01:09:01.861965 Display::i_handleDisplayResize: uScreenId=0 pvVRAM=00007f8d84000000 w=800 h=600 bpp=32 cbLine=0xC80 flags=0x1 origin=0,0
1240 03:39:24.558347 AsyncCompletion: Task 0x007f8e182ba9c0 completed after 11 seconds

It does seem some hard disk trouble. I would investigate the host logs for
that. Run the long smart test.

- --
Cheers
Carlos E. R.

(from openSUSE 15.1 (Legolas))

[David C. Rankin]

Can someone give me some hints/tips/suggestions please?
Best regards,
George...

I have multiple versions of Leap created in VirtualBox without any problems. I
don't know about tricks or tips, but for VMs I only use ext4 as the
filesystem. (the default Leap/Tumbleweed will attempt to install with btrfs) I
generally allocate 2G RAM for the VMs (you can get by with as little as 512K)
and I set a variable amount of drive space to max at 8G - I install with the
"minimal X" pattern and then add the KDE3 repo for a full KDE3 install, as
well as kernel source and all the gcc/g++ tools and a number of libraries.

The only thing I do is download the net-iso, point virtualbox to the .iso file
to install and then just follow the prompts but make sure in partitioning to
wipe the suggested default and create a / with ext4 -- no need for a separate
/ and /home on the VM for my use.

That's about all I can tell you. I'll load 15.2 in the next week or so and if
I see anything different, then I'll report back.

[George R Goffe]

David,

Thanks for responding.

I have used a LOT of the options for the install but this is NOT new for me.

I just tried a "production" version of VB and it failed as well. Same result.

I'm looking at /var/log/messages. It seems that oom-reaper was invoked due to VB. I'm beginning to think that something in my config is breaking the install or rather causing a break.

Is there a way to get all the config parameters dumped to a file? Like I said, my config is complicated.

Maybe I'll try Tumbleweed 13x.

Best regards,

George...

Can someone give me some hints/tips/suggestions please?
Best regards,
George...

I have multiple versions of Leap created in VirtualBox without any problems. I
don't know about tricks or tips, but for VMs I only use ext4 as the
filesystem. (the default Leap/Tumbleweed will attempt to install with btrfs) I
generally allocate 2G RAM for the VMs (you can get by with as little as 512K)
and I set a variable amount of drive space to max at 8G - I install with the
"minimal X" pattern and then add the KDE3 repo for a full KDE3 install, as
well as kernel source and all the gcc/g++ tools and a number of libraries.

The only thing I do is download the net-iso, point virtualbox to the .iso file
to install and then just follow the prompts but make sure in partitioning to
wipe the suggested default and create a / with ext4 -- no need for a separate
/ and /home on the VM for my use.

That's about all I can tell you. I'll load 15.2 in the next week or so and if
I see anything different, then I'll report back.
--
David C. Rankin, J.D.,P.E.