-
Notifications
You must be signed in to change notification settings - Fork 175
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
vic-init fails with 4 NICs configured #2802
Comments
This is easy to recreate on a local ESX env - simply create enough vSwitches that each network can be assigned it's own vNIC. Deploying with --debug=1 inhibits the rebooting, but to get hold of the logs requires modifying |
Related: #1674 |
Correcting the slot number in the configuration doesn't help:
It should be:
So it looks like the slot to pci logic is flawed as well. |
If the fix is simple we should fix it, but if not then can we document our way out of this? |
@karthik-narayan we might need inputs on this |
closing as dup of #1674 |
Reopening for actual fix |
For 0.8.0 we will limit the number of nics to 3. This will need to get fixed before we ship a 1.0. |
When this is fixed, we must also fix #3125. |
Copied the note that appears numerous times in the core doc into the release notes for 0.9:
|
@stuclem looks good. |
Thanks @mlh78750 |
Debt label because this derives from the use of pcislot identifiers which are not predictable past 3 NICs. We should be using the /label content - should also have a vsphere enhancement to allow specification of label so as not to rely on config spec ordering. |
We were generating 192, 224, 256 and 288 as the PCI slot number for a 4 NIC system. The problem was 288 was behind pciBridge8 which we don't have in our VM. We are now starting from 1184 (1184, 1216, 1248, and 1280...) to get the persistent PCI slot mapping between hypervisor and guest. Fixes #2802
We were generating 192, 224, 256 and 288 as the PCI slot number for a 4 NIC system. The problem was 288 was behind pciBridge8 which we don't have in our VM. We are now starting from 1184 (1184, 1216, 1248, and 1280...) to get the persistent PCI slot mapping between hypervisor and guest. Fixes vmware#2802
The output in #2794 is caused by vic-init exiting when 4 separate NICs are specified. The installer generates the wrong slot number for the 4th NIC and vic-init does not handle this well:
We should not be exiting vic-init if/when this occurs - while rebooting the VM can redress some issues, a hardware misconfiguration is not one of them. Without vic-init running we cannot enable SSH or console for manual debugging.
vic-init output for reference:
vic-machine log:
PCI devices with label's
slot data for ethernet vNICs from vmx:
The text was updated successfully, but these errors were encountered: