[one-users] Configure hosts to use qemu without kvm hardware acceleration
Hi all, I'm building a testing environment and I need to use some kvm virtual machines as host nodes. Since I cannot enable the hardware acceleration inside a virtual machine*, I'd like to know how to configure a host node like the configuration that there is in the kvm sandbox (that actually use qemu-kvm with the parameter -no-kvm). As frontend node and host nodes I'm using Debian 7, and opennebula is installed from the .deb packages for Debian 7 downloaded from the opennebula website. Thank you all, Marco Fanti *I know it is possible to enable nested kvm inside kvm, the guide is http://palexster.wordpress.com/2013/03/06/enable-nested-virtualization-on-kvm/ , BUT this features has a lot of bug, especially in the old versions of kvm like the one in Debian Stable, so I'd like a more stable solution, like to use qemu without kvm as in the kvm sandbox ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] ceph+flashcache datastore driver
On 24/03/14 17:57, Stuart Longland wrote: Well, I'm starting to have a stab at creating a driver to do local caching on hosts for RBD storage. For those who want to follow progress, I've thrown a repository up here: http://git.longlandclan.yi.org/?p=opennebula-ceph-flashcache.git;a=summary A heads up on the progress, I now have some basic cloning/linking of images working. Our cluster was pushing about 120MB/sec prior to me rebuilding the compute nodes and embarking on this cache experiment. The set up here is: The Ceph nodes are combined MON/OSD hosts with two OSDs and a MON each. The machines each have two 1Gbps NICs, one of which connects to an isolated unmanaged 1Gbps switch, the other connects to a second unmanaged 1Gbps switch shared with the VM hosts. The VM hosts themselves have 3 1Gbps NICs, one as a VLAN trunk for VMs, the other two are bonded to form a 2Gbps link. They each have a 60GB SSD, of which 8GB is taken up by the OS. The rest is LVM. For testing I set up a VM with 100% of its disk cached to local SSD. The initial read was slow, about 20MB/sec (dependent on overall cluster/network load), however, subsequent reads is showing speeds of approximately 218MB/sec. So once the cache is filled up, there's a big jump in read performance. -- Stuart Longland Contractor _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration
Shouldn't qemu fallback to paravitualization when hardware acceleration isn't available? http://blog.vmsplice.net/2011/03/should-i-use-qemu-or-kvm.html?m=1 I was confident that qemu could detect on the fly if hardware virtualization was available and would enable it or not as appropriate. until your post... Are you getting an error with OpenNebula? Thanks, Jon A On Mar 25, 2014 1:41 AM, Marco Fanti marco.fa...@studenti.polito.it wrote: Hi all, I'm building a testing environment and I need to use some kvm virtual machines as host nodes. Since I cannot enable the hardware acceleration inside a virtual machine*, I'd like to know how to configure a host node like the configuration that there is in the kvm sandbox (that actually use qemu-kvm with the parameter -no-kvm). As frontend node and host nodes I'm using Debian 7, and opennebula is installed from the .deb packages for Debian 7 downloaded from the opennebula website. Thank you all, Marco Fanti *I know it is possible to enable nested kvm inside kvm, the guide is http://palexster.wordpress.com/2013/03/06/enable-nested-virtualization-on-kvm/ , BUT this features has a lot of bug, especially in the old versions of kvm like the one in Debian Stable, so I'd like a more stable solution, like to use qemu without kvm as in the kvm sandbox ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration
It is nevertheless an interesting feature. For example, when emulating legacy OSes like DOS or Windows 3.11 you have to disable KVM anyway, or the OS does not boot. (And yes, we *do* have sometimes to bring legacy operating systems into OpenNebula :-)) Maybe we can add it as a feature? It should be mainly a matter of passing the -no-kvm switch cheers, Carlo Daffara - Messaggio originale - Da: Jon three1...@gmail.com A: Marco Fanti marco.fa...@studenti.polito.it Cc: Users OpenNebula users@lists.opennebula.org Inviato: Martedì, 25 marzo 2014 10:20:50 Oggetto: Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration Shouldn't qemu fallback to paravitualization when hardware acceleration isn't available? http://blog.vmsplice.net/2011/03/should-i-use-qemu-or-kvm.html?m=1 I was confident that qemu could detect on the fly if hardware virtualization was available and would enable it or not as appropriate. until your post... Are you getting an error with OpenNebula? Thanks, Jon A On Mar 25, 2014 1:41 AM, Marco Fanti marco.fa...@studenti.polito.it wrote: Hi all, I'm building a testing environment and I need to use some kvm virtual machines as host nodes. Since I cannot enable the hardware acceleration inside a virtual machine*, I'd like to know how to configure a host node like the configuration that there is in the kvm sandbox (that actually use qemu-kvm with the parameter -no-kvm). As frontend node and host nodes I'm using Debian 7, and opennebula is installed from the .deb packages for Debian 7 downloaded from the opennebula website. Thank you all, Marco Fanti *I know it is possible to enable nested kvm inside kvm, the guide is http://palexster.wordpress.com/2013/03/06/enable-nested-virtualization-on-kvm/ , BUT this features has a lot of bug, especially in the old versions of kvm like the one in Debian Stable, so I'd like a more stable solution, like to use qemu without kvm as in the kvm sandbox ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
[one-users] Virtual Machine keeps failing
Hello, I know that making templates is the hardest part of opennebula and I searched and tried as much as I can. So I need your help please... I want to setup any Virtual Machine (it does not even matter if it is a Linux, Ubuntu, Debian, CentOS, Windows ) I use OpenNebula 4.4.1. with Xen 4.2.4 When I try to run a VM it fails and this is the log file: Can anyone help me with a good solid tutorial or something? Tue Mar 25 11:59:38 2014 [DiM][I]: New VM state is CLEANUP. Tue Mar 25 11:59:40 2014 [DiM][I]: New VM state is PENDING Tue Mar 25 11:59:51 2014 [DiM][I]: New VM state is ACTIVE. Tue Mar 25 11:59:51 2014 [LCM][I]: New VM state is PROLOG. Tue Mar 25 12:00:37 2014 [LCM][I]: New VM state is BOOT Tue Mar 25 12:00:37 2014 [VMM][I]: Generating deployment file: /var/lib/one/vms/4/deployment.1 Tue Mar 25 12:00:37 2014 [VMM][I]: ExitCode: 0 Tue Mar 25 12:00:37 2014 [VMM][I]: Successfully execute network driver operation: pre. Tue Mar 25 12:00:38 2014 [VMM][I]: Command execution fail: cat EOT | /var/tmp/one/vmm/xen4/deploy '/var/lib/one//datastores/0/4/deployment.1' 'rackserver7' 4 rackserver7 Tue Mar 25 12:00:38 2014 [VMM][I]: libxl: error: libxl_utils.c:273:libxl_read_file_contents: failed to open /var/lib/one//datastores/0/4/deployment.1: Permission denied Tue Mar 25 12:00:38 2014 [VMM][I]: Failed to read config file: /var/lib/one//datastores/0/4/deployment.1: Permission denied Tue Mar 25 12:00:38 2014 [VMM][E]: Unable Tue Mar 25 12:00:38 2014 [VMM][I]: ExitCode: 3 Tue Mar 25 12:00:38 2014 [VMM][I]: Failed to execute virtualization driver operation: deploy. Tue Mar 25 12:00:38 2014 [VMM][E]: Error deploying virtual machine: Unable Tue Mar 25 12:00:38 2014 [DiM][I]: New VM state is FAILED -- Kind regards, Duez Christophe Student at University of Antwerp : Master of Industrial Sciences: Electronics-ICT E christophe.d...@student.uantwperen.be L linkedin duez-christophehttp://www.linkedin.com/pub/duez-christophe/74/7/39 ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration
OpenNebula can be configured to use qemu emulation instead of kvm virtualization but not per VM but in the driver definition. The type of the VM driver can be changed from kvm to qemu in oned.conf. For example: --8-- VM_MAD = [ name = kvm, executable = one_vmm_ssh, arguments = -t 15 -r 0 kvm, default= vmm_ssh/vmm_ssh_kvm.conf, type = qemu ] --8-- This is used in the tryout images as they are already virtualized. It will be also useful for the infrastructure Marco Fanti wants to create. Still this is not useful to have mixed virtualized and emulated machines in the same host. Maybe this parameter can be changed to a VM setting instead of a driver setting. On Tue, Mar 25, 2014 at 10:28 AM, Carlo Daffara carlo.daff...@cloudweavers.eu wrote: It is nevertheless an interesting feature. For example, when emulating legacy OSes like DOS or Windows 3.11 you have to disable KVM anyway, or the OS does not boot. (And yes, we *do* have sometimes to bring legacy operating systems into OpenNebula :-)) Maybe we can add it as a feature? It should be mainly a matter of passing the -no-kvm switch cheers, Carlo Daffara - Messaggio originale - Da: Jon three1...@gmail.com A: Marco Fanti marco.fa...@studenti.polito.it Cc: Users OpenNebula users@lists.opennebula.org Inviato: Martedì, 25 marzo 2014 10:20:50 Oggetto: Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration Shouldn't qemu fallback to paravitualization when hardware acceleration isn't available? http://blog.vmsplice.net/2011/03/should-i-use-qemu-or-kvm.html?m=1 I was confident that qemu could detect on the fly if hardware virtualization was available and would enable it or not as appropriate. until your post... Are you getting an error with OpenNebula? Thanks, Jon A On Mar 25, 2014 1:41 AM, Marco Fanti marco.fa...@studenti.polito.it wrote: Hi all, I'm building a testing environment and I need to use some kvm virtual machines as host nodes. Since I cannot enable the hardware acceleration inside a virtual machine*, I'd like to know how to configure a host node like the configuration that there is in the kvm sandbox (that actually use qemu-kvm with the parameter -no-kvm). As frontend node and host nodes I'm using Debian 7, and opennebula is installed from the .deb packages for Debian 7 downloaded from the opennebula website. Thank you all, Marco Fanti *I know it is possible to enable nested kvm inside kvm, the guide is http://palexster.wordpress.com/2013/03/06/enable-nested-virtualization-on-kvm/ , BUT this features has a lot of bug, especially in the old versions of kvm like the one in Debian Stable, so I'd like a more stable solution, like to use qemu without kvm as in the kvm sandbox ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration
Thank you for the answers, this was the log file of the vm that don't start because of the missing kvm module (so it does not fall back to qemu --no-kvm when the kvm module is not present on the system. Mon Mar 24 15:11:02 2014 [DiM][I]: New VM state is ACTIVE. Mon Mar 24 15:11:02 2014 [LCM][I]: New VM state is PROLOG. Mon Mar 24 15:11:02 2014 [VM][I]: Virtual Machine has no context Mon Mar 24 15:11:03 2014 [LCM][I]: New VM state is BOOT Mon Mar 24 15:11:03 2014 [VMM][I]: Generating deployment file: /var/lib/one/vms/2/deployment.0 Mon Mar 24 15:11:03 2014 [VMM][I]: ExitCode: 0 Mon Mar 24 15:11:03 2014 [VMM][I]: Successfully execute network driver operation: pre. Mon Mar 24 15:11:03 2014 [VMM][I]: Command execution fail: cat EOT | /var/tmp/one/vmm/kvm/deploy '/var/lib/one//datastores/0/2/deployment.0' 'on-compute1' 2 on-compute1 Mon Mar 24 15:11:03 2014 [VMM][I]: error: Failed to create domain from /var/lib/one//datastores/0/2/deployment.0 Mon Mar 24 15:11:03 2014 [VMM][I]: error: internal error Child process (LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin HOME=/ /usr/bin/kvm -no-user-config -device ? -device pci-assign,? -device virtio-blk-pci,? -device virtio-net-pci,? -device scsi-disk,?) unexpected exit status 1: Could not access KVM kernel module: No such file or directory Mon Mar 24 15:11:03 2014 [VMM][I]: failed to initialize KVM: No such file or directory Mon Mar 24 15:11:03 2014 [VMM][I]: No accelerator found! Mon Mar 24 15:11:03 2014 [VMM][I]: Mon Mar 24 15:11:03 2014 [VMM][E]: Could not create domain from /var/lib/one//datastores/0/2/deployment.0 Mon Mar 24 15:11:03 2014 [VMM][I]: ExitCode: 255 2014-03-25 13:14 GMT+01:00 Javier Fontan jfon...@opennebula.org: OpenNebula can be configured to use qemu emulation instead of kvm virtualization but not per VM but in the driver definition. The type of the VM driver can be changed from kvm to qemu in oned.conf. For example: --8-- VM_MAD = [ name = kvm, executable = one_vmm_ssh, arguments = -t 15 -r 0 kvm, default= vmm_ssh/vmm_ssh_kvm.conf, type = qemu ] --8-- This is used in the tryout images as they are already virtualized. It will be also useful for the infrastructure Marco Fanti wants to create. Still this is not useful to have mixed virtualized and emulated machines in the same host. Maybe this parameter can be changed to a VM setting instead of a driver setting. Thank you Javier, this solution is actually documented in the oned.conf comments, I'm sorry that i didn't noticed it before. On Tue, Mar 25, 2014 at 10:28 AM, Carlo Daffara carlo.daff...@cloudweavers.eu wrote: It is nevertheless an interesting feature. For example, when emulating legacy OSes like DOS or Windows 3.11 you have to disable KVM anyway, or the OS does not boot. (And yes, we *do* have sometimes to bring legacy operating systems into OpenNebula :-)) Maybe we can add it as a feature? It should be mainly a matter of passing the -no-kvm switch cheers, Carlo Daffara - Messaggio originale - Da: Jon three1...@gmail.com A: Marco Fanti marco.fa...@studenti.polito.it Cc: Users OpenNebula users@lists.opennebula.org Inviato: Martedì, 25 marzo 2014 10:20:50 Oggetto: Re: [one-users] Configure hosts to use qemu without kvm hardware acceleration Shouldn't qemu fallback to paravitualization when hardware acceleration isn't available? http://blog.vmsplice.net/2011/03/should-i-use-qemu-or-kvm.html?m=1 I was confident that qemu could detect on the fly if hardware virtualization was available and would enable it or not as appropriate. until your post... Are you getting an error with OpenNebula? Thanks, Jon A On Mar 25, 2014 1:41 AM, Marco Fanti marco.fa...@studenti.polito.it wrote: Hi all, I'm building a testing environment and I need to use some kvm virtual machines as host nodes. Since I cannot enable the hardware acceleration inside a virtual machine*, I'd like to know how to configure a host node like the configuration that there is in the kvm sandbox (that actually use qemu-kvm with the parameter -no-kvm). As frontend node and host nodes I'm using Debian 7, and opennebula is installed from the .deb packages for Debian 7 downloaded from the opennebula website. Thank you all, Marco Fanti *I know it is possible to enable nested kvm inside kvm, the guide is http://palexster.wordpress.com/2013/03/06/enable-nested-virtualization-on-kvm/, BUT this features has a lot of bug, especially in the old versions of kvm like the one in Debian Stable, so I'd like a more stable solution, like to use qemu without kvm as in the kvm sandbox ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list
[one-users] Ceph image creation works ... but disk is empty (ONE 4.6 beta)
Hi List, I've setup a 5 node ceph cluster and 2 kvm nodes. Everything is working fine. Ceph cluster is healthy, etc. I'm able to list images on the frontend, kvm nodes. Datastore is added succesfully: DATASTORE 100 INFORMATION ID : 100 NAME : ceph_one_ds USER : oneadmin GROUP : oneadmin CLUSTER: - TYPE : IMAGE DS_MAD : ceph TM_MAD : ceph BASE PATH : /var/lib/one//datastores/100 DISK_TYPE : RBD DATASTORE CAPACITY TOTAL: : 1T FREE: : 967.1G USED: : 42.1G LIMIT: : - PERMISSIONS OWNER : um- GROUP : u-- OTHER : --- DATASTORE TEMPLATE BASE_PATH=/var/lib/one//datastores/ BRIDGE_LIST=oned1 CEPH_HOST=ceph1 ceph2 ceph3 CEPH_SECRET=e6ba4d0a-0e20-4a3a-9053-0a7ef50a02b9 CEPH_USER=libvirt CLONE_TARGET=SELF DISK_TYPE=RBD DS_MAD=ceph LN_TARGET=NONE POOL_NAME=one TM_MAD=ceph TYPE=IMAGE_DS IMAGES 8 9 However, If I import an image from the marketplace it gets registered succesfully. But actually the image is empty: qemu-img info rbd:one/one-9 image: rbd:one/one-9 file format: raw virtual size: 0 (0 bytes) disk size: unavailable Image details: IMAGE 9 INFORMATION ID : 9 NAME : centos_ceph_test USER : oneadmin GROUP : oneadmin DATASTORE : ceph_one_ds TYPE : OS REGISTER TIME : 03/24 10:47:10 PERSISTENT : Yes SOURCE : one/one-9 PATH : http://marketplace.c12g.com/appliance/4fc76a938fb81d351702/download/0 SIZE : 10G STATE : rdy RUNNING_VMS: 0 PERMISSIONS OWNER : um- GROUP : --- OTHER : --- IMAGE TEMPLATE DEV_PREFIX=hd MD5=94176fbba17f5efcabb0b05fc55b1a16 I've manually created a qemu-image (qemu-img convert -f qcow2 -O raw debian_squeeze.qcow2 rbd:one/debian2) and made a libvirt xml based on openenbula default. That image just works. Everything seems to be in order, no errors anywhere, but result in empty images nevertheless. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Native GlusterFS support
Maybe I have a bug in the documentation. Set the ds mad to fs and check if it works. On Mon, Mar 24, 2014 at 9:03 PM, Stefan Kooman ste...@bit.nl wrote: Quoting Javier Fontan (jfon...@opennebula.org): Now that the packages for OpenNebula 4.6 beta are ready is anyone willing to give a shot to the gluster integration? Any feedback is welcome. Post: http://opennebula.org/native-glusterfs-image-access-for-kvm-drivers/ Packages: http://opennebula.org/software/ Documentation: http://docs.opennebula.org/4.6/administration/storage/gluster_ds.html I'm trying glusterfs on Ubuntu Saucy (frontend) and Ubuntu Trusty (nodes). I've followed the documentation but something is not working. Might be me missing something here. DATASTORE 101 INFORMATION ID : 101 NAME : gluster_gv0_ds USER : oneadmin GROUP : oneadmin CLUSTER: - TYPE : IMAGE DS_MAD : shared TM_MAD : shared BASE PATH : /var/lib/one//datastores/101 DISK_TYPE : DATASTORE CAPACITY TOTAL: : 0M FREE: : 0M USED: : 0M LIMIT: : - PERMISSIONS OWNER : um- GROUP : u-- OTHER : --- DATASTORE TEMPLATE BASE_PATH=/var/lib/one//datastores/ CLONE_TARGET=SYSTEM DISK_TYPE=GLUSTER DS_MAD=shared GLUSTER_HOST=gluster1:24007 GLUSTER_VOLUME=gv0 LN_TARGET=NONE TM_MAD=shared TYPE=IMAGE_DS oneadmin@oned1:~$ onedatastore list ID NAMESIZE AVAIL CLUSTER IMAGES TYPE DS TM 0 system215.9G 92% - 0 sys -shared 1 default 215.9G 92% - 1 img fs shared 2 files 17.6G 38% - 0 fil fs ssh 100 ceph_one_ds 1T 91% - 2 img ceph ceph 101 gluster_gv0_d 0M - - 0 img shared shared It reports 0M - AVAIL. The system and default datastores are mounted glusterfs volume (so glusterfs does work on frontend/nodes). If I try to import an image from marketplace I get the following error: [ImageAllocate] Cannot determine Image SIZE. Datastore driver 'shared' not available. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
[one-users] Opennebula 4.4 plus dual stack IPv4/IPv6
Does anyone have a worked example of an OpenNebula virtual network that has a so-called dual stack setup, i.e with both ipv4 and ipv6 addresses created and assigned by OpenNebula in a fixed network? The documents appear to indicate that you can do only ipv4 or ipv6, not both. Steve -- Steven C. Timm, Ph.D (630) 840-8525 t...@fnal.gov http://home.fnal.gov/~timm/ Fermilab Scientific Computing Division, Scientific Computing Services Quad. Grid and Cloud Services Dept., Associate Dept. Head for Cloud Computing ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Native GlusterFS support
Quoting Javier Fontan (jfon...@opennebula.org): Maybe I have a bug in the documentation. Set the ds mad to fs and check if it works. I changed the setting and that helped :). Unfortunately qemu logs an error: qemu-system-x86_64: -drive file=gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4,if=none,id=drive-virtio-disk0,format=raw,cache=none: Gluster connection failed for server=gluster1 port=24007 volume=gv0 image=53f35f5ca22540d558404caa3b9907d4 transport=tcp ... while the gluster volume is mounted correctly on this hypervisor I'll try to debug this later. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Opennebula 4.4 plus dual stack IPv4/IPv6
Quoting Steven Timm (t...@fnal.gov): Does anyone have a worked example of an OpenNebula virtual network that has a so-called dual stack setup, i.e with both ipv4 and ipv6 addresses created and assigned by OpenNebula in a fixed network? The documents appear to indicate that you can do only ipv4 or ipv6, not both. You can do both, in a FIXED as well as a RANGED network. You do need to enable CONTEXT_FORCE_IPV4, see here my template: VIRTUAL NETWORK TEMPLATE CLUSTER=kvm_cluster CONTEXT_FORCE_IPV4=yes DNS= GATEWAY= GATEWAY6= NETWORK_ADDRESS= NETWORK_MASK= Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Native GlusterFS support
It could be a permission problem. Try this command with both root and oneadmin: $ qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 If just oneadmin does not work you may need to change some configuration parameters to let non-root users access glusterfs: http://docs.opennebula.org/4.6/administration/storage/gluster_ds.html#glusterfs-volume-configuration On Tue, Mar 25, 2014 at 4:52 PM, Stefan Kooman ste...@bit.nl wrote: Quoting Javier Fontan (jfon...@opennebula.org): Maybe I have a bug in the documentation. Set the ds mad to fs and check if it works. I changed the setting and that helped :). Unfortunately qemu logs an error: qemu-system-x86_64: -drive file=gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4,if=none,id=drive-virtio-disk0,format=raw,cache=none: Gluster connection failed for server=gluster1 port=24007 volume=gv0 image=53f35f5ca22540d558404caa3b9907d4 transport=tcp ... while the gluster volume is mounted correctly on this hypervisor I'll try to debug this later. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl -- Javier Fontán Muiños Developer OpenNebula - The Open Source Toolkit for Data Center Virtualization www.OpenNebula.org | @OpenNebula | github.com/jfontan ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Native GlusterFS support
Quoting Javier Fontan (jfon...@opennebula.org): It could be a permission problem. Try this command with both root and oneadmin: $ qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 qemu-img: Could not open 'gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4': Unknown protocol It turns out that stock libvirt/qemu in Ubuntu Trusty do _not_ have gluster support. I've installed a PPA [1], with that PPA installed it does work: qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 image: gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 file format: raw virtual size: 40M (41943040 bytes) disk size: 40M (ttylinux image). I'll go ahead and install the PPA on the other node too. Thanks for helping me out. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] Hosts in error state.
I dint modified that file and these are the permissions: root@ubuntu:/var/lib/one/remotes/datastore/fs# ls -l total 19 -rwxr-xr-x 1 oneadmin oneadmin 2488 Feb 3 20:42 clone -rwxr-xr-x 1 oneadmin oneadmin 3700 Feb 3 20:42 cp -rwxr-xr-x 1 oneadmin oneadmin 3032 Feb 3 20:42 mkfs -rwxr-xr-x 1 oneadmin oneadmin 2474 Feb 3 20:42 monitor -rwxr-xr-x 1 oneadmin oneadmin 2532 Feb 3 20:42 rm -rwxr-xr-x 1 oneadmin oneadmin 2433 Feb 3 20:42 stat From: jme...@opennebula.org Date: Mon, 24 Mar 2014 15:21:00 +0100 Subject: Re: [one-users] Hosts in error state. To: meduri.jagade...@msn.com CC: users@lists.opennebula.org Hi, you are getting sh: 1: /var/lib/one/remotes/datastore/fs/monitor: Permission denied have you modified that file? what are the permissions / owner of that file? they should belong to oneadmin. ls -l of that file would help. cheers,Jaime On Tue, Mar 18, 2014 at 8:00 PM, Meduri Jagadeesh meduri.jagade...@msn.com wrote: Wed Mar 19 00:16:04 2014 [ReM][D]: Req:3136 UID:0 VirtualMachinePoolInfo result SUCCESS, VM_POOL/VM_POOL Wed Mar 19 00:16:04 2014 [ReM][D]: Req:6816 UID:0 VirtualMachinePoolInfo invoked, -2, -1, -1, -1 Wed Mar 19 00:16:04 2014 [ReM][D]: Req:6816 UID:0 VirtualMachinePoolInfo result SUCCESS, VM_POOL/VM_POOL Wed Mar 19 00:16:05 2014 [ReM][D]: Req:2352 UID:0 VirtualMachinePoolInfo invoked, -2, -1, -1, -1 Wed Mar 19 00:16:05 2014 [ReM][D]: Req:2352 UID:0 VirtualMachinePoolInfo result SUCCESS, VM_POOL/VM_POOL Wed Mar 19 00:16:05 2014 [InM][D]: Monitoring host localhost (32) Wed Mar 19 00:16:05 2014 [ReM][D]: Req:2352 UID:0 VirtualMachinePoolInfo invoked, -2, -1, -1, -1 Wed Mar 19 00:16:05 2014 [InM][D]: Monitoring datastore default (1) Wed Mar 19 00:16:06 2014 [ImM][I]: Command execution fail: /var/lib/one/remotes/datastore/fs/monitor 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 1 Wed Mar 19 00:16:06 2014 [ImM][I]: sh: 1: /var/lib/one/remotes/datastore/fs/monitor: Permission denied Wed Mar 19 00:16:06 2014 [ImM][I]: ExitCode: 126 Wed Mar 19 00:16:06 2014 [ImM][E]: Error monitoring datastore 1: - Wed Mar 19 00:16:07 2014 [ReM][D]: Req:2352 UID:0 VirtualMachinePoolInfo result SUCCESS, VM_POOL/VM_POOL Wed Mar 19 00:16:07 2014 [InM][D]: Monitoring datastore files (2) Wed Mar 19 00:16:07 2014 [ImM][I]: Command execution fail: /var/lib/one/remotes/datastore/fs/monitor PERTX0RSSVZFUl9BQ1RJT05fREFUQT48REFUQVNUT1JFPjxJRD4yPC9JRD48VUlEPjA8L1VJRD48R0lEPjA8L0dJRD48VU5BTUU+b25lYWRtaW48L1VOQU1FPjxHTkFNRT5vbmVhZG1pbjwvR05BTUU+PE5BTUU+ZmlsZXM8L05BTUU+PFBFUk1JU1NJT05TPjxPV05FUl9VPjE8L09XTkVSX1U+PE9XTkVSX00+MTwvT1dORVJfTT48T1dORVJfQT4xPC9PV05FUl9BPjxHUk9VUF9VPjE8L0dST1VQX1U+PEdST1VQX00+MDwvR1JPVVBfTT48R1JPVVBfQT4wPC9HUk9VUF9BPjxPVEhFUl9VPjE8L09USEVSX1U+PE9USEVSX00+MDwvT1RIRVJfTT48T1RIRVJfQT4wPC9PVEhFUl9BPjwvUEVSTUlTU0lPTlM+PERTX01BRD5mczwvRFNfTUFEPjxUTV9NQUQ+c3NoPC9UTV9NQUQ+PEJBU0VfUEFUSD4vdmFyL2xpYi9vbmUvL2RhdGFzdG9yZXMvMjwvQkFTRV9QQVRIPjxUWVBFPjI8L1RZUEU+PERJU0tfVFlQRT4wPC9ESVNLX1RZUEU+PENMVVNURVJfSUQ+LTE8L0NMVVNURVJfSUQ+PENMVVNURVI+PC9DTFVTVEVSPjxUT1RBTF9NQj4xNzE4NTwvVE9UQUxfTUI+PEZSRUVfTUI+MTMyODg8L0ZSRUVfTUI+PFVTRURfTUI+MTwvVVNFRF9NQj48SU1BR0VTPjwvSU1BR0VTPjxURU1QTEFURT48Q0xPTkVfVEFSR0VUPjwhW0NEQVRBW1NZU1RFTV1dPjwvQ0xPTkVfVEFSR0VUPjxEU19NQUQ+PCFbQ0RBVEFbZnNdXT48L0RTX01BRD48TE5fVEFSR0VUPjwhW0NEQVRBW1NZU1RFTV1dPjwvTE5fVEFSR0VUPjxUTV9NQUQ+PCFbQ0RBVEFbc3NoXV0+PC9UTV9NQUQ+PFRZUEU+PCFbQ0RBVEFbRklMRV9EU11dPjwvVFlQRT48L1RFTVBMQVRFPjwvREFUQVNUT1JFPjwvRFNfRFJJVkVSX0FDVElPTl9EQVRBPg== 2 Wed Mar 19 00:16:07 2014 [ImM][I]: sh: 1: /var/lib/one/remotes/datastore/fs/monitor: Permission denied Wed Mar 19 00:16:07 2014 [ImM][I]: ExitCode: 126 Wed Mar 19 00:16:07 2014 [ImM][E]: Error monitoring datastore 2: - Wed Mar 19 00:16:08 2014 [ReM][D]: Req:1488 UID:0 VirtualMachinePoolInfo invoked, -2, -1, -1, -1 Wed Mar 19 00:16:08 2014
Re: [one-users] Native GlusterFS support
Can you share that ppa for the community please? I am assuming the qemu-img ppa that has gluster support. FYI the fedora releases 19 and 20 do have glusterfs support in qemu without extra packages. On Tue, Mar 25, 2014 at 1:09 PM, Stefan Kooman ste...@bit.nl wrote: Quoting Javier Fontan (jfon...@opennebula.org): It could be a permission problem. Try this command with both root and oneadmin: $ qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 qemu-img: Could not open 'gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4': Unknown protocol It turns out that stock libvirt/qemu in Ubuntu Trusty do _not_ have gluster support. I've installed a PPA [1], with that PPA installed it does work: qemu-img info gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 image: gluster://gluster1:24007/gv0/53f35f5ca22540d558404caa3b9907d4 file format: raw virtual size: 40M (41943040 bytes) disk size: 40M (ttylinux image). I'll go ahead and install the PPA on the other node too. Thanks for helping me out. Gr. Stefan -- | BIT BV http://www.bit.nl/Kamer van Koophandel 09090351 | GPG: 0xD14839C6 +31 318 648 688 / i...@bit.nl ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
Re: [one-users] ceph+flashcache datastore driver
Try bcache as a flash backend, I feel its more flexible as a caching tier and its well integrated into the kernel. The kernel 3.10.X version is now quite mature so an epel6 long term kernel would work great. We are using it in a linux based production SAN as a cache tier with pci-e SSDs, a very flexible subsystem and rock solid. On Tue, Mar 25, 2014 at 5:03 AM, Stuart Longland stua...@vrt.com.au wrote: On 24/03/14 17:57, Stuart Longland wrote: Well, I'm starting to have a stab at creating a driver to do local caching on hosts for RBD storage. For those who want to follow progress, I've thrown a repository up here: http://git.longlandclan.yi.org/?p=opennebula-ceph-flashcache.git;a=summary A heads up on the progress, I now have some basic cloning/linking of images working. Our cluster was pushing about 120MB/sec prior to me rebuilding the compute nodes and embarking on this cache experiment. The set up here is: The Ceph nodes are combined MON/OSD hosts with two OSDs and a MON each. The machines each have two 1Gbps NICs, one of which connects to an isolated unmanaged 1Gbps switch, the other connects to a second unmanaged 1Gbps switch shared with the VM hosts. The VM hosts themselves have 3 1Gbps NICs, one as a VLAN trunk for VMs, the other two are bonded to form a 2Gbps link. They each have a 60GB SSD, of which 8GB is taken up by the OS. The rest is LVM. For testing I set up a VM with 100% of its disk cached to local SSD. The initial read was slow, about 20MB/sec (dependent on overall cluster/network load), however, subsequent reads is showing speeds of approximately 218MB/sec. So once the cache is filled up, there's a big jump in read performance. -- Stuart Longland Contractor _ ___ \ /|_) | T: +61 7 3535 9619 \/ | \ | 38b Douglas StreetF: +61 7 3535 9699 SYSTEMSMilton QLD 4064 http://www.vrt.com.au ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org ___ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
