Red Hat Bugzilla – Bug 578129
Install fails with greater than 3072 Mb of guest RAM
Last modified: 2011-08-08 10:42:52 EDT
Description of problem: Impossible to install guest with virt-install on a guest with more than 3072 RAM.
Version-Release number of selected component (if applicable): python-virtinst-0.400.3-7.el5
How reproducible: Everytime.
virt-install --os-type=linux --os-variant=virtio26 --accelerate --location=ftp://... --disk path=/dev/mapper/... --name=mac3 --ram=3073 --bridge=br_prod --vnc -x "ks=ftp://.../kickstart.cfg ip=... netmask=... gateway=..."
--os-variant=rhel5 does not change anything
Steps to Reproduce:
1. The hypervisor (HP ProLiant BL460c G1) has 8Gb of RAM + 8 Gb of swap, and no virtual machine installed.
2. The threshold of RAM attributed to the guest in order to have the error is 3072
checking if image is initramfs... it isn't (invalid compressed format (err=1)); looks like an initrd
Freeing initrd memory: 7226k freed
or with --ram=4096 :
checking if image is initramfs... it isn't (bad gzip magic numbers); looks like an initrd
Freeing initrd memory: 7226k freed
EXT2-fs: unable to read superblock
isofs_fill_super: bread failed, dev=md1, iso_blknum=16, block=32
No filesystem could mount root, tried: ext2 iso9660
Kernel panic – not syncing: VFS: Unable to mount root fs on unknown-block(9,1)
With --ram=3072 or less, the installation is working perfectly.
The same problem is seen on the other identical server. But when the RAM of the server is 16 Gb on another server, the problem is not seen.
Interesting, but this probably a KVM or kernel issue. Reassigning to KVM.
This request was evaluated by Red Hat Product Management for
inclusion in the current release of Red Hat Enterprise Linux.
Because the affected component is not scheduled to be updated in the
current release, Red Hat is unfortunately unable to address this
request at this time. Red Hat invites you to ask your support
representative to propose this request, if appropriate and relevant,
in the next release of Red Hat Enterprise Linux.
This request was erroneously denied for the current release of
Red Hat Enterprise Linux. The error has been fixed and this
request has been re-proposed for the current release.
The problem complaint here is a very common use case (>3GB RWM)
and this problem has not been found elsewhere. To further
investigate this issue we need far more detail of the test
conditions where this was observed. It would also be helpful
for the reporter to setup this failure scenario on a remote
machine for a developer to examine.
Bobby, we have never been able to reproduce this bug internally. However, Prarit did recently find a TLB flushing bug on x86 that could explain your issue.
I have made test kernels with that patch available on http://people.redhat.com/riel/.bz578129/
This is a bit of a long shot, since we have not reproduced the bug internally, but the bug matches your symptoms. Please let us know whether the test kernel resolves your issue, or whether we should keep looking.
Tried to reproduce this issue using kvm directly with kvm-83-231.el5,failed to reproduce.
I tired to install guest with -m 4G and -m 6G,my host memory is 8G.both can install successful.
I will try to install system using virt-install later,update soon
Tried using virt-install,failed to reproduce.
#virt-install --os-type=linux --os-variant=virtio26 --accelerate -l ftp://download.englab.nay.redhat.com/pub/rhel/rel-eng/latest-RHEL-6/6.1/Server/x86_64/os/ --disk path=/var/lib/libvirt/images/foo.img,size=6 --name=foo --ram=4096 --bridge=switch --vnc -x "ks=http://home.englab.nay.redhat.com/~nzhang/kickstart/ftp/ks-rhel6u1-x86_64.cfg" --debug
Wed, 27 Apr 2011 13:03:11 DEBUG Requesting libvirt URI default
Wed, 27 Apr 2011 13:03:11 DEBUG Received libvirt URI 'qemu:///system'
Wed, 27 Apr 2011 13:03:11 DEBUG Requesting virt method 'default'
Wed, 27 Apr 2011 13:03:11 DEBUG Received virt method 'hvm'
Wed, 27 Apr 2011 13:03:11 DEBUG Hypervisor name is 'kvm'
Wed, 27 Apr 2011 13:03:11 DEBUG parse_disk: returning ('/var/lib/libvirt/images/foo.img', None, None, 'disk', None, False, False, 6.0, True, None)
Wed, 27 Apr 2011 13:03:11 DEBUG Detected storage as type 'file'
Wed, 27 Apr 2011 13:03:11 DEBUG DistroInstaller location is a network source.
Wed, 27 Apr 2011 13:03:23 DEBUG Attempting to detect distro:
Wed, 27 Apr 2011 13:03:23 DEBUG Fetching URI: ftp://download.englab.nay.redhat.com/pub/rhel/rel-eng/latest-RHEL-6/6.1/Server/x86_64/os/.treeinfo
Wed, 27 Apr 2011 13:03:23 DEBUG Saved file to /var/lib/libvirt/boot/virtinst-.treeinfo.8yxXzr
Retrieving file .treeinfo... | 1.7 kB 00:00
Wed, 27 Apr 2011 13:03:23 DEBUG Fetching URI: ftp://download.englab.nay.redhat.com/pub/rhel/rel-eng/latest-RHEL-6/6.1/Server/x86_64/os/images/pxeboot/vmlinuz
Wed, 27 Apr 2011 13:03:23 DEBUG Saved file to /var/lib/libvirt/boot/virtinst-vmlinuz.oFz2zv===========================- ] 0.0 B/s | 3.6 MB --:-- ETA
Retrieving file vmlinuz...
I tried reproduce this issue using qemu-kvm directly and virt-install tool.all failed.
Did you still can reproduce this issue on rhel5.7? would you please have a try comment12 patch? thanks.
Host kernel version:
Several people tried to reproduce it, and failed. It was reported once, a long time ago, probably with RHEL5.4. This is typical for bugs that we close for 5.8.