cloudstack-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Engelmann Florian <florian.engelm...@everyware.ch>
Subject Re: Ubuntu 16.04, Openvswitch networking issue
Date Mon, 06 Mar 2017 10:34:29 GMT
Hi,

looks like I was wrong. The typo did not fix our issue. We are still not sure whats going
on but we are not able to deploy virtual router on xenserver if 

 system.vm.default.hyper

is empty.

All the best,
Florian

________________________________________
From: Engelmann Florian <florian.engelmann@everyware.ch>
Sent: Wednesday, March 1, 2017 9:53 AM
To: users@cloudstack.apache.org
Subject: Re: Ubuntu 16.04, Openvswitch networking issue

Dear Rafael,

my apologies! We had a typo in "router.template.xenserver".

All the best,
Florian


________________________________________
From: Rafael Weingärtner <rafaelweingartner@gmail.com>
Sent: Tuesday, February 28, 2017 4:08 PM
To: users@cloudstack.apache.org
Subject: Re: Ubuntu 16.04, Openvswitch networking issue

Did you configure the parameter “system.vm.default.hypervisor”? If so, ACS
will try to deploy only on hosts of the hypervisor type configured there.

If the parameter is not configured, ACS will get the first it finds. After
it deployed the VR (let’s say on KVM host of cluster "x"), if a redeploy is
needed, it will retry to do so on the same cluster "x". This is a guess now
(have to check the code to make sure), if there are no more available hosts
on the cluster "x" it first deployed the VR, it will try other available
clusters.

On Tue, Feb 28, 2017 at 6:44 AM, Engelmann Florian <
florian.engelmann@everyware.ch> wrote:

> Dear Rafael,
>
> thank you very much. I am wondering why just KVM is tried in our case? The
> Zone got 3 clusters, one 3 node Xenserver 6.5, one 1 node Xenserver 7.0 and
> one 3 node KVM cluster.
>
> Is there some Setting we have to edit?
>
> All the best,
> Florian
>
>
>
> ________________________________________
> From: Rafael Weingärtner <rafaelweingartner@gmail.com>
> Sent: Wednesday, February 22, 2017 5:00 PM
> To: users@cloudstack.apache.org
> Subject: Re: Ubuntu 16.04, Openvswitch networking issue
>
> Yes, I am able to explain. I am still not getting the whole picture here.
> VR = virtual router
>
> First condition:
>
> > if (dest.getCluster() != null) {
> >
>
>
>    - If we are forcing a VR to a specific cluster, normally this happens
>    when we are starting a router that was already deployed. If it is a new
> VR
>    “dest.getCluster()” will return null( I think this is your case).
>    - Then, we go for the “else” part of the code. In this block, we check
>    if there is a default hypervisor configured,
>    “_resourceMgr.getDefaultHypervisor”.
>    - If there is not a default hypervisor, we load
>    “_resourceMgr.getSupportedHypervisorTypes” all supported hypervisor in
> the
>    given zone. And this will be the value of “hypervisors”.
>    - After all that, at the method
>    “com.cloud.network.router.NetworkHelperImpl.deployRouter(
> RouterDeploymentDefinition,
>    boolean)”, we get the first hypervisor available and try to deploy a
> router
>    in on of the available hosts for this hypervisor type.
>
>
> Was that clear for you?
>
>
> On Wed, Feb 22, 2017 at 9:14 AM, Engelmann Florian <
> florian.engelmann@everyware.ch> wrote:
>
> > Hi,
> >
> > after digging into the source code I wonder if
> >
> > https://github.com/apache/cloudstack/blob/master/server/
> > src/com/cloud/network/router/NetworkHelperImpl.java
> >
> > line 453
> > final List<HypervisorType> hypervisors = getHypervisors(
> > routerDeploymentDefinition);
> >
> > line 458
> >  for (final Iterator<HypervisorType> iter = hypervisors.iterator();
> > iter.hasNext();) {
> >
> > ever gets more than one item in the list? In our case it looks like there
> > is only one item in this list (KVM) even if we got three clusters - one
> > with KVM and two with Xenserver and the VM gets deployed on Xenserver.
> >
> > Are you able to explain the following function to me?
> >
> >     protected List<HypervisorType> getHypervisors(final
> > RouterDeploymentDefinition routerDeploymentDefinition) throws
> > InsufficientServerCapacityException {
> >         final DeployDestination dest = routerDeploymentDefinition.
> > getDest();
> >         List<HypervisorType> hypervisors = new
> ArrayList<HypervisorType>();
> >         if (dest.getCluster() != null) {
> >             if (dest.getCluster().getHypervisorType() ==
> > HypervisorType.Ovm) {
> >                 hypervisors.add(getClusterToStartDomainRouterF
> > orOvm(dest.getCluster().getPodId()));
> >             } else {
> >                 hypervisors.add(dest.getCluster().getHypervisorType());
> >             }
> >         } else {
> >             final HypervisorType defaults = _resourceMgr.
> > getDefaultHypervisor(dest.getDataCenter().getId());
> >             if (defaults != HypervisorType.None) {
> >                 hypervisors.add(defaults);
> >             } else {
> >                 // if there is no default hypervisor, get it from the
> > cluster
> >                 hypervisors = _resourceMgr.getSupportedHypervisorTypes(
> dest.getDataCenter().getId(),
> > true, routerDeploymentDefinition.getPlan().getPodId());
> >             }
> >         }
> >
> > How will that function return a list?
> >
> > All the best,
> > Florian
> >
> > ________________________________________
> > From: Engelmann Florian <florian.engelmann@everyware.ch>
> > Sent: Wednesday, February 22, 2017 12:24 PM
> > To: users@cloudstack.apache.org
> > Subject: Re: Ubuntu 16.04, Openvswitch networking issue
> >
> > Hi,
> >
> > this behavior was caused by the following setting:
> >
> > system.vm.default.hypervisor = Xenserver
> >
> > After setting this global parameter to NULL we are able to deploy
> > SystemVMs (virtual router) on KVM.
> >
> > Sad to say, we are no more able to deploy SystemVMs on Xenserver since we
> > changed the parameter?!?!
> >
> > Allocating the VR with id=4213 in datacenter com.cloud.dc.DataCenterVO$$
> > EnhancerByCGLIB$$6fda4e4f@2 with the hypervisor type KVM
> > [...]
> > VM state transitted from :Starting to Stopped with event:
> > OperationFailedvm's original host id: null new host id: null host id
> before
> > state transition: null
> >
> >
> > Is anybody able to point me in the direction where to find the logic (on
> > Github) behind how ACS decides which Hypervisor should be used to deploy
> a
> > VirtualRouter SystemVM? Why is KVM chosen?
> >
> > All the best,
> > Florian
> >
> >
> >
> > ________________________________________
> > From: Engelmann Florian
> > Sent: Monday, February 20, 2017 5:25 PM
> > To: users@cloudstack.apache.org
> > Subject: Re: Ubuntu 16.04, Openvswitch networking issue
> >
> > Hi Rafael,
> >
> > We do use the following setup (Test environment currently):
> >
> > ACS 4.9.2
> > 1x Xenserver 6.5 Cluster (3 nodes)
> > 1x Xenserver 7.0 Cluster (1 node)
> > 1x Ubuntu 16.04 KVM Cluster (3 nodes)
> >
> > Networking = Advanced Zone VPC and Virtual Router
> >
> > I noticed the error Message was just informational an not the real
> > problem. The problem we got is:
> >
> > [...]
> > Allocating the VR with id=4185 in datacenter com.cloud.dc.DataCenterVO$$
> > EnhancerByCGLIB$$caa6c375@2 with the hypervisor type XenServer
> > [...]
> > Cluster: 6 has HyperVisorType that does not match the VM, skipping this
> > cluster
> > [...]
> >
> > We tried to force ACS to use the KVM systemVM template but for some
> reason
> > ACS refuses to use that System offering.
> >
> > Hosts:
> > Name        ewcstack-vh023-test
> > Host Tags="kvm"
> >
> > Primary storage:
> > Name: ewcstack-vh023-test,Local Storage: Storage Tags="vol-local-kvm"
> >
> > System offering:
> > Name  custom-local-sm-kvm
> > Storage Tags="vol-local-kvm"
> > Host Tags="kvm"
> >
> > Network offering:
> > Name        custom local kvm
> > System offering="custom-local-sm-kvm"
> >
> > Disk offering:
> > Name        custom local kvm
> > Storage Tags="vol-local-kvm"
> >
> >
> > Creating a Instance with a network offering "custom-local-sm-kvm" doesn't
> > stop ACS from using a XenServer systemVM template. Why?
> >
> > All the best,
> > Florian
> >
> >
> > ________________________________________
> > From: Rafael Weingärtner <rafaelweingartner@gmail.com>
> > Sent: Friday, February 17, 2017 4:08 PM
> > To: users@cloudstack.apache.org
> > Subject: Re: Ubuntu 16.04, Openvswitch networking issue
> >
> > I think we may need more information. ACS version, network deployment
> type,
> > and hypervisors?
> >
> > On Fri, Feb 17, 2017 at 10:02 AM, Engelmann Florian <
> > florian.engelmann@everyware.ch> wrote:
> >
> > > Hi,
> > >
> > > sorry I ment "I am NOT able to solve"....
> > >
> > > ________________________________________
> > > From: Engelmann Florian <florian.engelmann@everyware.ch>
> > > Sent: Friday, February 17, 2017 3:36 PM
> > > To: users@cloudstack.apache.org
> > > Subject: Ubuntu 16.04, Openvswitch networking issue
> > >
> > > Hi,
> > >
> > > another error I am able to solve:
> > >
> > > 2017-02-17 15:24:36,097 DEBUG [c.c.a.ApiServlet] (catalina-exec-26:ctx-
> > > 30020483) (logid:d303f8ef) ===START===  192.168.252.76 -- GET
> > > command=createNetwork&response=json&zoneId=e683eeaa-
> > > 92c9-4651-91b9-165939f9000c&name=net-kvm008&displayText=
> > > net-kvm008&networkOf
> > > 2017-02-17 15:24:36,135 DEBUG [c.c.n.g.BigSwitchBcfGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network, the physical isolation type is not BCF_SEGMENT
> > > 2017-02-17 15:24:36,136 DEBUG [o.a.c.n.c.m.ContrailGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network
> > > 2017-02-17 15:24:36,137 DEBUG [c.c.a.m.DirectAgentAttache]
> > > (DirectAgent-144:ctx-b2cdad73) (logid:eb129204) Seq
> > > 179-6955246674520311671: Response Received:
> > > 2017-02-17 15:24:36,137 DEBUG [c.c.a.t.Request] (StatsCollector-5:ctx-
> > 4298a591)
> > > (logid:eb129204) Seq 179-6955246674520311671: Received:  { Ans: ,
> MgmtId:
> > > 345049101620, via: 179(ewcstack-vh003-test), Ver: v1, Flags: 10, {
> > > GetStorageStatsAnswer } }
> > > 2017-02-17 15:24:36,137 DEBUG [c.c.n.g.MidoNetGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) design
> > called
> > > 2017-02-17 15:24:36,138 DEBUG [c.c.h.o.r.Ovm3HypervisorGuru]
> > > (StatsCollector-5:ctx-4298a591) (logid:eb129204)
> > > getCommandHostDelegation: class com.cloud.agent.api.
> > GetStorageStatsCommand
> > > 2017-02-17 15:24:36,138 DEBUG [c.c.h.XenServerGuru]
> > (StatsCollector-5:ctx-4298a591)
> > > (logid:eb129204) getCommandHostDelegation: class com.cloud.agent.api.
> > > GetStorageStatsCommand
> > > 2017-02-17 15:24:36,139 DEBUG [c.c.n.g.MidoNetGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network, the physical isolation type is not MIDO
> > > 2017-02-17 15:24:36,139 DEBUG [c.c.a.m.DirectAgentAttache]
> > > (DirectAgent-72:ctx-656a03ae) (logid:dd7ada9e) Seq
> > 217-8596245788743434945:
> > > Executing request
> > > 2017-02-17 15:24:36,141 DEBUG [c.c.n.g.NiciraNvpGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network
> > > 2017-02-17 15:24:36,142 DEBUG [o.a.c.n.o.OpendaylightGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network
> > > 2017-02-17 15:24:36,144 DEBUG [c.c.n.g.OvsGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network
> > > 2017-02-17 15:24:36,163 DEBUG [o.a.c.n.g.SspGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) SSP not
> > > configured to be active
> > > 2017-02-17 15:24:36,164 DEBUG [c.c.n.g.BrocadeVcsGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design this network
> > > 2017-02-17 15:24:36,165 DEBUG [c.c.n.g.NuageVspGuestNetworkGuru]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Refusing
> to
> > > design network using network offering 54 on physical network 200
> > > 2017-02-17 15:24:36,166 DEBUG [o.a.c.e.o.NetworkOrchestrator]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) Releasing
> > > lock for Acct[3426fb73-70ad-47d9-9c5d-355f34891438-fen]
> > > 2017-02-17 15:24:36,188 DEBUG [c.c.a.ApiServlet]
> > > (catalina-exec-26:ctx-30020483 ctx-430b6ae1) (logid:d303f8ef) ===END===
> > > 192.168.252.76 -- GET  command=createNetwork&
> > > response=json&zoneId=e683eeaa-92c9-4651-91b9-165939f9000c&
> > > name=net-kvm008&displayText=net-kvm00
> > >
> > >
> > > We do not use BigSwitch or anything like this, just plain Openvswitch
> > with
> > > Ubuntu 16.04. Any idea whats going on?
> > >
> > > All the best,
> > > Florian
> > >
> > > EveryWare AG
> > > Florian Engelmann
> > > Systems Engineer
> > > Zurlindenstrasse 52a
> > > CH-8003 Zürich
> > >
> > > T  +41 44 466 60 00
> > > F  +41 44 466 60 10
> > >
> > > florian.engelmann@everyware.ch
> > > www.everyware.ch
> > >
> >
> >
> >
> > --
> > Rafael Weingärtner
> >
>
>
>
> --
> Rafael Weingärtner
>



--
Rafael Weingärtner

Mime
View raw message