Hi My Fault, you are right the cluster requirements are also used for system datastores. Note that the default cluster behaves as described for VNETs. This behavior is not coherent with the documentation as you mentioned in a previous mail, so it seems a bug. I filled an issue [1] to look at this before 4.12.
Thanks for the detailed feedback :) Cheers Ruben [1] http://dev.opennebula.org/issues/3597 On Sat, Feb 14, 2015 at 1:27 AM, Steven C Timm <t...@fnal.gov> wrote: > One more followup: > > host 156 + vnet2 + ds 100/102, all outside the cluster, no problem > host 156 + vnet2 + ds100/102, all in the cluster, no problem > > host 156 and vnet2 in the cluster, DS outside of the cluster, problem. > SCHED_MESSAGE="Fri Feb 13 18:06:29 2015 : No system datastore meets > SCHED_DS_REQUIREMENTS: CLUSTER_ID = 101 & !(PUBLIC_CLOUD = YES)" > > host 156 in the cluster, vnet2 and DS out of the cluster > No error message but it never matches either. > > Fri Feb 13 18:24:29 2015 [Z0][HOST][D]: Discovered Hosts (enabled): > 0 2 156 > Fri Feb 13 18:24:29 2015 [Z0][SCHED][D]: VM 1058: Host 0 filtered out. It > does not fulfill SCHED_REQUIREMENTS. > Fri Feb 13 18:24:29 2015 [Z0][SCHED][D]: VM 1058: Host 2 filtered out. It > does not fulfill SCHED_REQUIREMENTS. > Fri Feb 13 18:24:29 2015 [Z0][SCHED][I]: Scheduling Results: > Virtual Machine: 1058 > > PRI ID - HOSTS > ------------------------ > 1 156 > > PRI ID - DATASTORES > ------------------------ > 0 100 > 0 0 > > > Fri Feb 13 18:24:29 2015 [Z0][SCHED][I]: VM 1058: No suitable System DS found > for Host: 156. Filtering out host. > > Steve Timm > > > > ________________________________________ > From: Steven C Timm > Sent: Friday, February 13, 2015 6:01 PM > To: Ruben S. Montero > Cc: users@lists.opennebula.org > Subject: RE: [one-users] clusters in 4.8 > > PS--if there are other vm's still launched and running from the time when the > datastore used to be part of > a cluster, could that confuse anything? Do I have to restart oned to clear > anything up? > > Steve Timm > > ________________________________________ > From: Steven C Timm > Sent: Friday, February 13, 2015 5:56 PM > To: Ruben S. Montero > Cc: users@lists.opennebula.org > Subject: RE: [one-users] clusters in 4.8 > > OK here we go: > > VM in question is taking an image from image store 102 (currently in no > cluster), > vnet 0 "routable private" from cluster 100 "cloud worker" > also a number of hosts, including hosts # 0 and 2, also part of cluster > "cloud worker" > > VM stays pending for ever, hold reason is below.--it is requiring that > cluster ID has to be 100. > > Same image and same datastore and same vnet outside of the cluster, work just > fine. > > Seems like if I require any resource from the cluster, in this case a vnet, > then all resources > have to be in the cluster. Am I missing something? > > Steve Timm > > > > [root@fclheadgpvm01 one]# onevm show 1054 | more > VIRTUAL MACHINE 1054 INFORMATION > ID : 1054 > NAME : CLI_PRIV_SLF6Vanilla-1054 > USER : oneadmin > GROUP : oneadmin > STATE : PENDING > LCM_STATE : LCM_INIT > RESCHED : No > START TIME : 02/13 17:44:52 > END TIME : - > DEPLOY ID : - > > VIRTUAL MACHINE MONITORING > NET_RX : 0K > USED MEMORY : 0K > USED CPU : 0 > NET_TX : 0K > > PERMISSIONS > OWNER : um- > GROUP : --- > OTHER : --- > > VM DISKS > ID TARGET IMAGE TYPE SAVE SAVE_AS > 0 vda SLF6Vanilla file NO - > > VM NICS > ID NETWORK VLAN BRIDGE IP MAC > 0 routable-private no br1 10.128.1.9 54:52:00:02:0d:09 > > USER TEMPLATE > NPTYPE="NPERNLM" > SCHED_MESSAGE="Fri Feb 13 17:46:29 2015 : No system datastore meets > SCHED_DS_REQ > UIREMENTS: CLUSTER_ID = 100 & !(PUBLIC_CLOUD = YES)" > SCHED_RANK="FREE_MEM" > SCHED_REQUIREMENTS="HYPERVISOR=\"kvm\" & HOSTNAME=\"cloudworker*\"" > > VIRTUAL MACHINE TEMPLATE > AUTOMATIC_REQUIREMENTS="CLUSTER_ID = 100 & !(PUBLIC_CLOUD = YES)" > CONTEXT=[ > > CTX_USER="PFVTRVI+PElEPjA8L0lEPjxHSUQ+MDwvR0lEPjxHUk9VUFM+PElEPjA8L0lEPjwvR1JP > VVBTPjxHTkFNRT5vbmVhZG1pbjwvR05BTUU+PE5BTUU+b25lYWRtaW48L05BTUU+PFBBU1NXT1JEPi9E > Qz1jb20vREM9RGlnaUNlcnQtR3JpZC9PPU9wZW5cMjBTY2llbmNlXDIwR3JpZC9PVT1TZXJ2aWNlcy9D > Tj1mY2xoZWFkZ3B2bTAxLmZuYWwuZ292PC9QQVNTV09SRD48QVVUSF9EUklWRVI+eDUwOTwvQVVUSF9E > UklWRVI+PEVOQUJMRUQ+MTwvRU5BQkxFRD48VEVNUExBVEU+PFRPS0VOX1BBU1NXT1JEPjwhW0NEQVRB > [root@fclheadgpvm01 one]# onedatastore list > ID NAME SIZE AVAIL CLUSTER IMAGES TYPE DS TM > 0 system 0M - - 0 sys - shared > 1 default 21.2G 85% - 0 img fs shared > 2 files 21.2G 85% - 0 fil fs ssh > 100 localnode - - - 0 sys - ssh > 102 cloud_images 20T 75% - 2 img fs shared > [root@fclheadgpvm01 one]# onevnet list > ID USER GROUP NAME CLUSTER BRIDGE > LEASES > 0 oneadmin oneadmin routable-private cloudworke br1 > 8 > 2 oneadmin oneadmin DynamicIP - br0 > 13 > 3 oneadmin oneadmin StaticIP - br0 > 0 > [root@fclheadgpvm01 one]# onehost list | more > ID NAME CLUSTER RVM ALLOCATED_CPU ALLOCATED_MEM STAT > 0 cloudworker1200 cloudwork 4 400 / 800 (50%) 7.4G / 15.6G (47%) on > 1 cloudworker1201 cloudwork 0 - - off > 2 cloudworker1202 cloudwork 0 0 / 800 (0%) 0K / 15.6G (0%) on > ________________________________________ > From: Ruben S. Montero [rsmont...@opennebula.org] > Sent: Friday, February 13, 2015 4:49 PM > To: Steven C Timm > Cc: users@lists.opennebula.org > Subject: Re: [one-users] clusters in 4.8 > > Yes, you can do: > > Cluster A: Host_A0, Host_A1... + VNET_A0, VNET_A1... > Cluster B: HostB0, HostB1... + VNET_B0, VNET_B1... > Cluster Default: DS, DS_System > > Then a VM that uses VNET_A0 + DS would be scheduled to Cluster A. Note > that using VNET_A0 constrain resources from Cluster A + Cluster > Default. > > Cheers > > Ruben > > On Fri, Feb 13, 2015 at 10:42 PM, Steven C Timm <t...@fnal.gov> wrote: >> I know if I just take the vnet and the datastore out of the cluster, and >> have no clusters at all, then everything >> will work.. I was hoping to have a cluster structure of (host,vnet) pairings >> that could >> all share a common data store. However from the documentation, it looks >> like if >> your template requests any resource that is part of a cluster (vnet or image >> from datastore) >> then the scheduler will constrain you to resources that are part of that >> same cluster. >> >> Is that correct? >> >> Steve Timm >> >> ________________________________________ >> From: Ruben S. Montero [rsmont...@opennebula.org] >> Sent: Friday, February 13, 2015 3:11 PM >> To: Steven C Timm >> Cc: users@lists.opennebula.org >> Subject: Re: [one-users] clusters in 4.8 >> >> Hi >> >> If both clusters has access to the same datastores, just move them out >> of the first cluster. When a datastore or network is not assigned to >> any cluster (cluster default) OpenNebula assumes it can be used with >> any host (no matter in which cluster is set). >> >> BTW, although you do not needed for your use case, 4.12 will come with >> extended VDC support to create complex provision scenarios. Basically >> you can define generic "resource providers" that aggregate any >> resource (cluster, host, network, datastores) more here >> http://opennebula.org/4-12-features-virtual-data-center-redesign/ >> >> Cheers >> >> On Fri, Feb 13, 2015 at 6:37 PM, Steven Timm <t...@fnal.gov> wrote: >>> >>> I have had my one4.8 host up for a while with a single cluster >>> that has 150 hosts, one vnet, and a system and image datastore. >>> >>> I am now adding hosts from a different vnet. >>> want to make second host + vnet cluster but still use >>> the same system and image data stores. >>> >>> What's the right way to do that.. just remove the datastores >>> from the first cluster... they can't be in more than one >>> cluster at a time, can they? >>> >>> >>> Thanks for any suggestions. >>> >>> Steve Timm >>> >>> >>> ------------------------------------------------------------------ >>> Steven C. Timm, Ph.D (630) 840-8525 >>> t...@fnal.gov http://home.fnal.gov/~timm/ >>> Office: Wilson Hall room 804 >>> Fermilab Scientific Computing Division, >>> Scientific Computing Facilities Quadrant., >>> Experimental Computing Facilities Dept., >>> Project Lead for Virtual Facility Project. >>> >>> >>> _______________________________________________ >>> Users mailing list >>> Users@lists.opennebula.org >>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org >> >> >> >> -- >> Ruben S. Montero, PhD >> Project co-Lead and Chief Architect >> OpenNebula - Flexible Enterprise Cloud Made Simple >> www.OpenNebula.org | rsmont...@opennebula.org | @OpenNebula > > > > -- > Ruben S. Montero, PhD > Project co-Lead and Chief Architect > OpenNebula - Flexible Enterprise Cloud Made Simple > www.OpenNebula.org | rsmont...@opennebula.org | @OpenNebula -- Ruben S. Montero, PhD Project co-Lead and Chief Architect OpenNebula - Flexible Enterprise Cloud Made Simple www.OpenNebula.org | rsmont...@opennebula.org | @OpenNebula _______________________________________________ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
