I installed OpenNebula 4.12.3 to aarch64 base pine64 machine, after a little struggle I have created a node with command “onehost create pine1 -i kvm -v kvm -n dummy”. But in oned.log file, I get this error message and “onehost list” lists the host with retry status.
Thanks for the help.
KVM is not totally ready yet for any ARM architecture as far as I know. I recommend you to go with Operating System Level Virtualization using LXD. This was already done, you can find it here: http://clox.cloud. At this moment, the project only supports Raspberry Pi 3 single board computers, but thanks to the people from Pine64 we are finishing support for their SBCs. We are currently testing and hope to release it at mid-january.
If you can’t wait till then, you can use this guides (http://clox.cloud/docs/long_road/index.html) for manually deploying the project on RPi3 and try to do it with a Pine64, you will only need to make small changes.
I compiled the source code as stated in the http://clox.cloud/docs/long_road/frontend_installation.html page successfully. But when I add the host to the cluster an error occurs in the oned.conf file. And host status waiting in the INIT state.
Can someone clarify the error and help me?
Thanks.
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:8400 UID:0 one.zone.raftstatus invoked
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:8400 UID:0 one.zone.raftstatus result SUCCESS, "<SERVER_ID>-1<…"
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:9600 UID:0 one.vmpool.info invoked , -2, -1, -1, -1
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:9600 UID:0 one.vmpool.info result SUCCESS, "<VM_POOL></VM_POOL>"
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:9600 UID:0 one.vmpool.info invoked , -2, -1, -1, -1
Sun Jan 28 09:13:04 2018 [Z0][ReM][D]: Req:9600 UID:0 one.vmpool.info result SUCCESS, "<VM_POOL></VM_POOL>"
Sun Jan 28 09:13:12 2018 [Z0][MKP][D]: Monitoring marketplace OpenNebula Public (0)
Sun Jan 28 09:13:13 2018 [Z0][InM][D]: Monitoring datastore default (1)
Sun Jan 28 09:13:13 2018 [Z0][InM][D]: Monitoring datastore files (2)
Sun Jan 28 09:13:13 2018 [Z0][MKP][D]: Marketplace OpenNebula Public (0) successfully monitored.
Sun Jan 28 09:13:14 2018 [Z0][ImM][D]: Datastore files (2) successfully monitored.
Sun Jan 28 09:13:14 2018 [Z0][ImM][D]: Datastore default (1) successfully monitored.
Sun Jan 28 09:13:14 2018 [Z0][ONE][E]: Error parsing host information: syntax error, unexpected VARIABLE, expecting EQUAL or EQUAL_EMPTY at line 35, columns 511:513. Monitoring information:
ARCH=aarch64
MODELNAME="HYPERVISOR=kvm
TOTALCPU=400
CPUSPEED=0
TOTALMEMORY=2035752
USEDMEMORY=544164
FREEMEMORY=1491588
FREECPU=360
USEDCPU=40
NETRX=87725336
NETTX=11598462
DS_LOCATION_USED_MB=7752
DS_LOCATION_TOTAL_MB=14832
DS_LOCATION_FREE_MB=6899
DS = [
ID = 0,
USED_MB = 7752,
TOTAL_MB = 14832,
FREE_MB = 6899
]
DS = [
ID = 1,
USED_MB = 7752,
TOTAL_MB = 14832,
FREE_MB = 6899
]
DS = [
ID = 2,
USED_MB = 7752,
TOTAL_MB = 14832,
FREE_MB = 6899
]
HOSTNAME=pine1
VM_POLL=YES
VERSION=“4.12.3”
Opennebula’s KVM driver won’t work. IM driver needs a few modifications to be able to monitor the virtualization node, but VM driver will definitely not work. Remember that KVM, and generally speaking HVM support is not ready yet for although ARM’s engineers have done a great work with ARM Virtualization Extensions in ARM v8 and v8.1 and KVM and Xen are being modified to run there…sooo, soon we will see it.
I’ve also been working to get virtualization working on a Pine64, and have had success with virt-manager/virsh using a custom compiled Armbian kernel with KVM enabled.
Working through lots of challenges with OpenNebula, the error message @czd is encountering can be fixed with a modification to the “cpu.sh” script in /usr/lib/one/remotes/im/kvm-probes.d/ on the head end. This issue is an unmatched open quote due to a grep statement not finding an expected string in /proc/cpuinfo.
To get around it, I changed this line:
grep -m 1 “Hardware” /proc/cpuinfo | cut -d: -f2 | sed -e ‘s/^ *//’ | sed -e ‘s/$/"/’
To this:
grep -m 1 “model name\|Hardware” /proc/cpuinfo | cut -d: -f2 | sed -e ‘s/^ *//’ | sed -e ‘s/$/"/’
Then do a “onehost sync --rsync --force” (I have better luck using the rsync method with my Pine64), and the monitoring script should no longer error out this way.
I currently have two Pine64’s in registered in my OpenNebula cloud now, although I am not sure I’ll be able to create any VM’s on them. I tried to create a new template, but the only Architecture options were i686 or x86_64.
Can OpenNebula create a VM with an aarch64 architecture?
