rancher-users #harvester

Channels

harvester

clever-queen-50249

05/20/2022, 2:47 AM

I have installed harvester 1.02 on 3 servers OVHCloud. Hardware 3 x AMD Epyc 7413 256GB Mem 2xNic I set the VIP to the public nic IP address Each server has a public nic and has a public IP assigned and is exposed. The second nic is private connecting all 3 servers via a vlan. On a vm I create, how do I assign a public IP, or expose ports publicly?

numerous-barista-78936

05/20/2022, 3:40 AM

Backing up hundreds of gigs of VMs to start over 😢 . Wish I could afford a dev environment

numerous-barista-78936

05/20/2022, 3:58 AM

Does anyone know if there is a way to cancel the Harvester upgrade process when its stuck?

flaky-coat-75909

05/20/2022, 11:02 AM

I have newbie question Harvester can only be installed direclty on real machine ? (bare metal) installation on VM are not possible right?

bland-insurance-21543

05/21/2022, 9:32 AM

I am new to using Harvester. I understand that things like passthrough are possible with Kubevirt, but not exactly front and center in the UI. I think i know how to get it working. To get started, I see where I can add modules to a harvester install in =harvester.config=. Is there a place where I can add kernel parameters for grub in this yaml file? If not, how do i configure this on my system without having the kernel paramters scrubbed by Harvester updates?

bland-insurance-21543

05/21/2022, 9:37 AM

also, how much "virtual machine complexity" is added by running the VM's inside of a container? does this result in extra layers of translation of vm machine code? or is it still pretty "flat"? i.e. is it basically a container running as a process that spawns the qemu process? i would imagine that there is more overhead here than with proxmox, but as long as the container isn't doing much other than spawning a VM, that it's not adding too much overhead. my other questions are more important, but i'm just curious about this. I have an AMD 5950x and harvester is using 0.5% but has reserved about 25%. i plan on running a single node cluster. if i ran a multinode cluster, would this reserved CPU number drop for the non-primary nodes?

modern-postman-45751

05/23/2022, 10:37 AM

Hi! I've got a question: I'm tinkering with a harvester setup on my spare time, and I'm trying to see if I can somehow make this work in my home-lab setup. I have two servers, with 2x10G NIC's + 1G NIC. In my current setup(not using harvester) the 2x10G interfaces are direct-attach between the servers for communication between the hosts, while I use the 1G interfaces to have the VM's reach the rest of the network + a bunch of routing trickery to make it work. all in all it is rather yanky, and after spinning up harvester on a few VM's I have to say I like what I see so far. But I'm a networking guy that's tinkering with this on my spare time, so I'm not sure how I can shoehorn this into my current setup. Am I right to assume that the harvester management interface, is the backplane for traffic between VM's and hosts? so I could in theory set up the harvester management network to be a simple point-to-point link between these machines? And is there anything I can do to also make the hosts reachable from the rest of the network, without making things too janky?

most-forest-2997

05/25/2022, 2:31 AM

hello! I am trying to figure out how to reconfigure the networking on an installed node, particularly adding another bond. I was reading some of the cOS toolkit docs and found the configuration in

/oem

but it didn't seem to get re-applied on a reboot. am I totally in the wrong place or did I miss a step?

sticky-summer-13450

05/29/2022, 2:03 PM

Having upgraded from Harvester 1.0.0 to 1.0.1, I now have this VM which is stuck in the state "Starting" phase. The events from the VM say:

FailedMount	
Pod virt-launcher-kube004-kgc64
(combined from similar events): Unable to attach or mount volumes: unmounted volumes=[rootdisk], unattached volumes=[libvirt-runtime cloudinitdisk-ndata public ephemeral-disks container-disks hotplug-disks sockets cloudinitdisk-udata private rootdisk cdrom-disk vardisk]: timed out waiting for the condition

FailedAttachVolume	
Pod virt-launcher-kube004-kgc64
AttachVolume.Attach failed for volume "pvc-50ef97d3-065d-4a5f-b986-36294317c269" : rpc error: code = Aborted desc = volume pvc-50ef97d3-065d-4a5f-b986-36294317c269 is not ready for workloads

What is the best way for me to discover the issue, I'm guessing in Longhorn, and recover from this situation? Thanks :-)

sticky-summer-13450

05/29/2022, 8:09 PM

Is VM anti-affinity possible in Harvester?

polite-megabyte-63476

05/31/2022, 6:04 PM

So my rke2 issue seems to be that it's not creating the service creds for the cloud provider in

/etc/kubernetes/cloud-config

refined-hamburger-4576

05/31/2022, 8:34 PM

What cloud/hosting provider do people usually run Harvester on? Are there some providers on which it is particularly easy to run Harvester?

orange-cpu-47176

06/03/2022, 12:16 PM

Hi @team I have a question about Harvester. We have a Server Dell r620 with 32 core e 96GB Ram, actually we manage the VMs with ESXI 7. Can we install Harvester instead of esxi 7 or we have to install ISO harvester on a vm created by esxi? In the extreme, do we necessarily need esxi or can we completely replace it with Harvester? Thank you

stale-painting-80203

06/03/2022, 9:02 PM

With Harvester, I am blocked on couple of issues and would very much appreciate someone's help or suggestions: 1. Unable to reach Harvester created VM using it's IP from outside the host on which it's running 2. VM not being able to resolve an internal FQDN for my Rancher instance I have installed Harvester on a physical server which has only one interface with a static IP: 10.255.156.88/24, gateway and a private DNS IP. I am able ping the Harvester server using it's IP and it is able to ping the Rancher instance using FQDN. I create VMs, which seem to get a dynamic IP assigned and I can see the IP from Harvester console: 10.52.0.116. I am also able to ping this IP from the host on which it's running. From within the VM I see a different IP:10.0.2.2/24. The VM is able to ping external domains such as google.com, but not the Rancher instance (I get Temporary failure in name resolution )

adorable-tomato-51499

06/04/2022, 11:58 AM

Hi team I have a simple network question about harvester. I installed Harvester on a physical server (homelab) which has a home network 192.x.x.0/24 range. I can reach the web interface and I can create virtual machines. When I create a virtual machine (ubuntu) I can only choose

management Network

and choose between bridge or masquerading mode. Both of them supply some 10.52.x.x/32 cluster ip address which is not reachable from my home network, only within the cluster itself. I would suspect that bridge mode would do this, but it's not. How can I configure the network settings so I can reach that virtual machine with SSH. Essentially I want to give that virtual machine a 192.x.x.0/24 range address. I have a VLAN capable switch, but I don't use VLANs. DHCP is running in the 192.x.x.0/24 range

✅ 1

adventurous-oyster-5549

06/05/2022, 1:22 PM

Hi I am trying to create a VM in harvester with some configurations populated in cloud init file and it has exceeded 2048 bytes. I am unable to create vm. Any advise how to i overcome it

wide-garage-9465

06/06/2022, 12:29 PM

Hi there 👋 So I have Harvester v1.0.2 installed on 3 servers in a remote data center. Testing all the Harvester features, I enabled "Maintenance Mode" on one of the servers. I see the Node being set to

SchedulingDisabled

and the status moving to

Entering maintenance mode

, but that's it. Nothing else happens. I'm pretty sure that it should at least evict all VMs from that node, right? I enabled trace logging, but there's absolutely nothing in the logs 🤔 Any hint appreciated 🙂

stale-painting-80203

06/07/2022, 5:13 AM

I am trying to create a VM using ubuntu minimal cloud image. The machine starts, but remains stuck at

GRUB_FORCE_PATUUID set, attempting initrdless boot.

The server does have internet connectivity and I have setup the network in the same way as my other VM, which used a regular ubuntu server ISO and works. Is there additional config required for creating VMs from cloud image?

hundreds-waiter-30127

06/07/2022, 6:58 AM

Hey all just getting into harvester and trying it out for the first time. I’m curious about storage pools. I’ve got 4 SSDs and 4 HDDs attached to my node. I didn’t find any possibility to schedule VMs on a set of storage pools (e.g ssd-pool / hdd-pool). Is this a feature coming in or is there a custom way of configuring this? Thanks a lot

little-dress-13576

06/08/2022, 12:42 PM

Could harvester installer support this hardware configuration for the same physical host? using 2 drives (maybe smaller nvme drives) as software raid1 mirroring for k3s install, and then use a few other nvme drives as data pool? is this possible via GUI installer or pxe installer?

hundreds-waiter-30127

06/08/2022, 4:33 PM

Hey all! I was playing around harvester and seem to be stuck. I want to create a k8s cluster with the harvester nodedriver on a rancher installation (a VM inside harvester) . As soon as I want to create a new cluster, I always keep getting stuck at the node-provisioning. My Nodepool VMs are deployed on harvester but it seems like the cloud-init is not triggered. When I open the console of the k8s-nodes, i always just see the default installer UI. Tried with ubuntu22.04 and centos 7.2009. I had it running once, but then I had just 1 single harvester node. anyone else with this problem or can reproduce it? thx! ProvisionLog:

Waiting for viable init node

Some docker logs of rancher:

2022/06/08 16:25:23 [INFO] [mgmt-auth-prtb-controller] Creating role project-owner in namespace p-dttv8
2022/06/08 16:25:23 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:23 [INFO] [mgmt-auth-prtb-controller] Creating roleBinding for subject user-dxlfr with role project-owner in namespace p-9nwqb
2022/06/08 16:25:24 [INFO] [mgmt-auth-prtb-controller] Creating roleBinding for subject user-dxlfr with role admin in namespace p-dttv8
2022/06/08 16:25:24 [INFO] [mgmt-auth-prtb-controller] Creating roleBinding for subject user-dxlfr with role project-owner in namespace p-dttv8
2022/06/08 16:25:24 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:24 [INFO] [mgmt-cluster-rbac-delete] Updating cluster c-m-wm9fhmsw
2022/06/08 16:25:24 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:24 [ERROR] Reconciler error: Operation cannot be fulfilled on <http://machinedeployments.cluster.x-k8s.io|machinedeployments.cluster.x-k8s.io> "harvester-k8s001-master": the object has been modified; please apply your changes to the latest version and try again
2022/06/08 16:25:25 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:25 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:25 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:25 [ERROR] error syncing 'fleet-default/harvester-k8s001-master-33169394-62dds': handler machine-provision: no machine owner ref, requeuing
2022/06/08 16:25:25 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:26 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:26 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:27 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:28 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:25:28 [INFO] [planner] rkecluster fleet-default/harvester-k8s001: waiting: waiting for viable init node
2022/06/08 16:28:28 [ERROR] Unknown error: Operation cannot be fulfilled on <http://preferences.management.cattle.io|preferences.management.cattle.io> "last-visited": the object has been modified; please apply your changes to the latest version and try again

lemon-bear-35731

06/09/2022, 9:32 PM

Is it possible to install a HA harvester cluster with 2 nodes with ample resources and a 3rd small witness node that does not participate in Longhorn or run VMs? If so, how do you set up the install configuration to install a node without longhorn storage

flaky-coat-75909

06/11/2022, 5:23 PM

I have old laptop asus and when I'm installing harvester on it I have a issue with a NIC I'm receing error

NIC wlp3s0 is down

how can I workaround it?

worried-lighter-48308

06/12/2022, 11:49 PM

Hey there. I was attempting to install Harvester on a repurposed system earlier. During the install, I got to the network configuration screen. There, I set the host name, but when trying to move to the next config item (selecting the management NIC), the installer seeming resets to the screen asking me what type of install to use (create new cluster or joining existing). Any ideas what is going on or where I can look to try figure it out?

bumpy-portugal-40754

06/13/2022, 3:04 PM

What's the usual places to check when a node to be provisioned hangs in the state "`waiting for agent to check in and apply initial plan`"? The actual VM is running fine in Harvester, although I cannot login via console because there is no password set for any user.

numerous-barista-78936

06/14/2022, 6:34 AM

Heres my desperate last resort call for help haha. We had a power outage with no UPS yet, cluster stopped starting up. Kubectl is inaccessible since rke2 is stopping repeatedly. To nobody's surprise, it looks like the usual culprit, etcd database is corrupted. This is a 2 node cluster, what are my options for recovery? Really hoping I can pull the VMs off the machines still.

sparse-ghost-91718

06/14/2022, 6:00 PM

Hello, I am trying to configure link aggregation with tagged VLANs on each harvester node, what are the command lines used for that ?

acoustic-action-80049

06/16/2022, 4:59 AM

Hello, I was trying to find Harvester <> Kubernetes support matrix. I was not able to find any documentation around it. Can someone please help me with same or direct me towards any documented support matrix? TIA!

quick-waitress-21632

06/17/2022, 7:48 AM

Hi all, what is used to provision Harvester at scale? Which provisioning tool is suggested? cheers Dino

magnificent-apartment-80494

06/17/2022, 3:20 PM

Good day folks! I'm new to Harvester (and Rancher in general). Recently I've converted a mix host environment from ESXI with vSphere over to Harvester. Currently I have three hosts in a cluster but I'm holding off on the fourth for the moment. I'm trying to get Live Migration working but it's being rather difficult. At the moment I have a small VM (2 vCPU and 8GB RAM) and two large VMs (16 vCPU and 64GB RAM). The smaller one I've been able to Live Migrate but it's inconsistent (failed twice, worked once) and then large VMs refuse to migrate at all. The large VMs are sitting on matched hardware (64 threads, 128GB RAM each, two nodes). When I initiate a migration from one node to the other it marks the VM as appropriate, but the Migration tab in the VM details never updates and the only thing in the VM log through the UI I see is a message about sync-vmi a couple times then nothing until the VM indicates that the last migration failed. Management interface on these nodes is the 1GB NIC, while VLAN network is sitting on 10GB NIC. Each node has 1 1GB, one iDRAC, and two SFP+ NICs. When I had these configured under ESXI I had the 1GB as the management interface, one of the SFPs for vServices (vMotion, vSAN, etc.), and the last SFP for VM traffic but I haven't seen if there's a similar way to utilize the NICs under Harvester.

Title

magnificent-apartment-80494

06/17/2022, 3:20 PM

Here are the only two messages I ever end up seeing on the VM log:

{"component":"virt-launcher","level":"warning","msg":"Domain id=1 name='default_flux-node-01' uuid=75aac619-8cef-5c48-a3e8-bbd5c965fbc2 is tainted: custom-ga-command","pos":"qemuDomainObjTaintMsg:6333","subcomponent":"libvirt","thread":"45","timestamp":"2022-06-17T15:23:43.245000Z"}Fri, Jun 17 2022 12:23:46 pm{"component":"virt-launcher","kind":"","level":"info","msg":"Synced vmi","name":"flux-node-01","namespace":"default","pos":"server.go:190","timestamp":"2022-06-17T15:23:46.191118Z","uid":"52c59f58-340d-4207-9fba-4aaa0c4973f1"}Fri, Jun 17 2022 12:23:46 pm

View count: 27