When I add NSX controller, I can see it powered on, powered off and then deleted. This really confuses me. I guess the deploy process detected some network issue so it cancelled the deployment but I cannot figure out why. While it's powered on, I log on to the controller and test the network. Its IP is 10.1.1.101/24. The ESXi host it resides has IP 10.1.1.1. And there's a gateway 10.1.1.254 in the subnet. The strange thing is the controller can ping its ESXi host but it cannot ping the gateway. Please advise how I can deploy the controller successfully. Thanks.
Hello Bolw,
Please make sure you have enough resources available on the cluster you are trying to deploy to, when the system is deploying check if the Controller can be seen as a VM booting in the vcenter console.
Controllers must be able to communicate with:
- The NSX Manager
- vmk0 of all your ESXi hosts
- Each other
There are a couple main reasons for Controller getting deleted:
1) Network settings for the Controller are incorrect. Check IP Pool configuration in Network and Security -> NSX Managers -> [Your NSX Manager] -> Manage -> Grouping Objects -> IP Pools, see if all settings are correct. Then check if you're connecting your Controller to the correct dvPg.
2) It takes too long for a Controller to come up (if your system resources are constrained), and NSX Manager gives up waiting and deletes it. Try adding a DHCP server to the portgroup where you're connecting your Controllers. They will not use DHCP address in the end, but it will speed up their deployment time.
I saw it powered on. I even log on to the controller for some network test but soon it's being deleted.
DmitriK 撰写:
Controllers must be able to communicate with:
- The NSX Manager
- vmk0 of all your ESXi hosts
- Each other
I think that's the problem I cannot figure out. My network goes like this
NSX Manager: 10.1.3.1/24
NSX Controller: 10.1.1.101/24
ESXi that hosts the NSX Controller: 10.1.1.1/24
Gateway: 10.1.1.254/24, 10.1.3.254/24
NSX Controller can ping ESXi but fail to ping the gateway and hence cannot reach NSX Manager. I'm sure the IP and netmask are correct. Why can't it ping the gateway? Is there some kind of firewall setting built in?
The Controller needs to complete a few tasks, before NSX Manager would consider it ready:
- Deploy, power on, go through the guest OS customisation (which configuration of Controller's network settings)
- Join the controller cluster (first Controller joins its own IP)
- Connect to NSX Manager and report for duty
You will be able to log in into the Controller once it gets its IP address. This however doesn't mean that the Controller has fully completed its deployment, or is able to talk to the NSX Manager.
NSX Manager will delete the Controller if it doesn't report for duty within a set period of time, irrespective of what may have caused it.
Which device is configured with the IP addresses of 10.1.1.254/24 and 10.1.3.254/24? Is this a physical router, or something else?
Could you please confirm:
- Can you can ping 10.1.1.254 from your ESXi host?
- Can you can ping NSX Manager (10.1.3.1) from your ESXi host?
DmitriK 撰写:
Which device is configured with the IP addresses of 10.1.1.254/24 and 10.1.3.254/24? Is this a physical router, or something else?
Could you please confirm:
- Can you can ping 10.1.1.254 from your ESXi host?
- Can you can ping NSX Manager (10.1.3.1) from your ESXi host?
The gateway is a virtual machine with three interfaces and CentOS 5.5 installed.
ESXi can ping both the gateway and NSX Manager.
If you create another VM and connect it to the same dvPortgroup that you use for your Controller, can it ping:
- The controller
- ESXi host
- NSX Manager
Another thing to check:
- when Controller is being deployed, and you can ping the ESXi host but not the default GW from it:
- can you see ARP requests from the Controller on your CentOS gateway VM?
..and one more thing to try: could you please try to deploy your Controller to the same dvPg where NSX Manager is? You'll need to make a new IP Pool so that it uses 10.1.3.0/24 range, of course.
DmitriK 撰写:
If you create another VM and connect it to the same dvPortgroup that you use for your Controller, can it ping:
- The controller
- ESXi host
- NSX Manager
Another thing to check:
- when Controller is being deployed, and you can ping the ESXi host but not the default GW from it:
- can you see ARP requests from the Controller on your CentOS gateway VM?
The VM behaves the same as the controller. It can ping the controller, ESXi host but not GW (NSX Manager). I also found it cannot ping other ESXi host in the same port group. It can only ping the ESXi that it resides in.
From CentOS gateway, I could see ARP requests for both ESXi and gateway but I only saw ARP reply from gateway to the controller. Does the ESXi not receive the ARP reply for its guest?
DmitriK 撰写:
..and one more thing to try: could you please try to deploy your Controller to the same dvPg where NSX Manager is? You'll need to make a new IP Pool so that it uses 10.1.3.0/24 range, of course.
My vCenter and NSX Manager are not in the vDS. I thought it suffices to let them reachable to all ESXi hosts. Is that correct?
It looks like the problem you're experiencing lays with the way ESXi/DVS networking is configured in your environment, and is not specific to NSX. As I mentioned, Controller and NSX Manager rely on IP communication between themselves and ESXi hosts for correct operation, so this communication has to be working correctly before NSX can be installed and used.
Could you please cross-post to the VMTN vSphere community? That way there are more people will see your question, and hopefully help get it resolved.
> My vCenter and NSX Manager are not in the vDS. I thought it suffices to let them reachable to all ESXi hosts. Is that correct?
Correct. You don't have to connect your NSX Manager, management interfaces of your ESXi hosts, or Controllers to dvPortgroups; you can use standard VLAN-backed portgroups.
I am curious!
I use nested ESXi but I don't use VLAN. DmitriK is right. It's the problem of my switch setting. For nested ESXi, I should enable promiscuous mode on the standard switches of physical ESXi. I can add controller now. Thanks very much.
Hi Guys,
I am running ESXi hosts on Vmware Workstation. Below is my design
I have one Physical ESXi host -> Windows virtual Machine -> VMware workstation -> ESXi running as VM in Workstation
I am trying to deploy NSX controller. Everytime it fails at 95% with the error operation timeout. I really need your help to troubleshoot this issue and deploy NSX contollers. I am using internal only network for VMware Workstation VM's.
Please help.
Hello
Did you managed to solve this ?
please read my post, hope it help
https://roie9876.wordpress.com/2014/08/30/deploying-nsx-v-controller-disappear-from-vsphere-client/
Hi All,
I face a error when deploy NSX manager with vCenter 5.5 server webclient :
details is here : fatclimber's blog: VMware NSX Manager 6.0.5 install
I could bypass it with the C# client but still nice to know if anyone has the same issue to share
Also, any consolidated posts on NSX deploying issue? I Think this could speed up the process alot..
Cheers,
Eric
Hey there...ran into this issue and found it to be a Resource Issue...check here: http://anthonyspiteri.net/nsx-bites-nsx-controller-deployment-issues/