[Starlingx-discuss] Deployment Option (error: compute boot in loop)

[Hu, Yong]

Intel_iommu=on and immmu are for PCI PT and SR-IOV, as far as I know.
There are other 2, like VT-x and VT-d flags, to be enabled. Please have a look in BIOS.


From: Himanshu Goyal <himanshugoyal500 at gmail.com>
Date: Thursday, 24 January 2019 at 12:07 AM
To: "Hu, Yong" <yong.hu at intel.com>
Cc: "Alonso, Juan Carlos" <juan.carlos.alonso at intel.com>, "starlingx-discuss at lists.starlingx.io" <starlingx-discuss at lists.starlingx.io>
Subject: Re: [Starlingx-discuss] Deployment Option (error: compute boot in loop)

Hi Yong,

Yes, virtualization setting are enabled in BIOS.
Below are the dmesg logs for IOMMU


[    0.000000] Policy zone: Normal
[    0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-3.10.0-862.11.6.el7.36.tis.x86_64 root=UUID=8c9ba8fa-ecae-4d66-98ac-a77fb66faae2 ro security_profile=standard module_blacklist=integrity,ima audit=0 tboot=false crashkernel=auto biosdevname=0 console=ttyS0,115200 iommu=pt usbcore.autosuspend=-1 hugepagesz=1G hugepages=2 selinux=0 enforcing=0 nmi_watchdog=panic,1 softlockup_panic=1 intel_iommu=on user_namespace.enable=1 hugepagesz=2M hugepages=0 default_hugepagesz=2M isolcpus=1,2 rcu_nocbs=1-35 kthread_cpus=0 irqaffinity=0 nopti nospectre_v2
[    0.000000] audit: disabled (until reboot)
[    0.000000] DMAR: IOMMU enabled
[    0.000000] PID hash table entries: 4096 (order: 3, 32768 bytes)


Many Thanks,
Himanshu Goyal


On Wed, Jan 23, 2019 at 7:43 PM Hu, Yong <yong.hu at intel.com<mailto:yong.hu at intel.com>> wrote:
Hi Himanshu,
For compute/worker node, please make sure Virtualization settings (such as VT-X and VT-D) ENABLED in BIOS.
They are mandatory requirements for compute node.

Regards,
Yong

From: Himanshu Goyal <himanshugoyal500 at gmail.com<mailto:himanshugoyal500 at gmail.com>>
Date: Wednesday, 23 January 2019 at 6:33 PM
To: "Hu, Yong" <yong.hu at intel.com<mailto:yong.hu at intel.com>>
Cc: "Alonso, Juan Carlos" <juan.carlos.alonso at intel.com<mailto:juan.carlos.alonso at intel.com>>, "starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>" <starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>>
Subject: Re: [Starlingx-discuss] Deployment Option (error: compute boot in loop)

Thanks a lot Juan and Yong,

Able to see compute host in "system-host-list" after connecting with a hub. starlingX  Compute installation has been done.
But after unlocking the compute nodes the compute node come into an endless boot loop.

dmesg log shows below error:

[ 20.967137] iTCO_vendor_support: vendor-support=0 [ 20.968785] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11 [ 20.968840] iTCO_wdt: unable to reset NO_REBOOT flag, device disabled by hardware/BIOS [ 21.066607] device-mapper: uevent: version 1.0.3 [ 21.067124] device-mapper: ioctl: 4.37.1-ioctl (2018-04-03) initialised: dm-devel at redhat.com<mailto:dm-devel at redhat.com>

Checked the BIOS setting those are as mentioned in installation document.

I'm using ISO Image available at below path:
http://mirror.starlingx.cengn.ca/mirror/starlingx/centos/2018.10/20181110/outputs/iso/


Regards,
Himanshu Goyal





On Tue, Jan 22, 2019 at 12:43 PM Hu, Yong <yong.hu at intel.com<mailto:yong.hu at intel.com>> wrote:
Himanshu,
Could you have a try with a hub which 2 mgt ports (from controller and compute) are plugged into?
Let’s assure the normal setup works first, and then figure out why the direct linkage of cable doesn’t work.

BTW: “worker” and “compute” are just different “personality” names in different STX version.
On your current setup “compute” will do, supposedly.
From: Himanshu Goyal <himanshugoyal500 at gmail.com<mailto:himanshugoyal500 at gmail.com>>
Date: Tuesday, 22 January 2019 at 12:32 AM
To: "Alonso, Juan Carlos" <juan.carlos.alonso at intel.com<mailto:juan.carlos.alonso at intel.com>>
Cc: "Hu, Yong" <yong.hu at intel.com<mailto:yong.hu at intel.com>>, "starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>" <starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>>
Subject: Re: [Starlingx-discuss] Deployment Option

Thanks Juan ,Yong

I tried both the commands output shows as below:

1) [wrsroot at controller-0 ~(keystone_admin)]$ system host-add -n compute-0 -p worker -m 00:1e:67:fd:3d:fe
usage: system host-add [-n <hostname>] [-p <personality>] [-s <subfunctions>]
                       [-m <mgmt_mac>] [-i <mgmt_ip>] [-I <bm_ip>]
                       [-T <bm_type>] [-U <bm_username>] [-P <bm_password>]
                       [-b <boot_device>] [-r <rootfs_device>]
                       [-o <install_output>] [-c <console>]
                       [-v <vsc_controllers>] [-l <location>]
                       [-D <true/false>]
system host-add: error: argument -p/--personality: invalid choice: 'worker' (choose from 'controller', 'compute', 'storage', 'network', 'profile')
[wrsroot at controller-0 ~(keystone_admin)]$
[wrsroot at controller-0 ~(keystone_admin)]$


2) [wrsroot at controller-0 ~(keystone_admin)]$ system host-add -n compute-0 -p compute -m 00:1e:67:fd:3d:fe
Host-add Rejected: Cannot add a compute host without specifying a mgmt_ip when static address allocation is configured.
[wrsroot at controller-0 ~(keystone_admin)]$


Regards,
Himanshu Goyal


On Mon, Jan 21, 2019 at 8:42 PM Alonso, Juan Carlos <juan.carlos.alonso at intel.com<mailto:juan.carlos.alonso at intel.com>> wrote:
Hi,

The personality of computes changed to “worker”, so the command should be:

system host-add -n compute-0 -p worker -m ${mac_address}

Regards.
Juan Carlos Alonso

From: Hu, Yong
Sent: Monday, January 21, 2019 8:51 AM
To: Himanshu Goyal <himanshugoyal500 at gmail.com<mailto:himanshugoyal500 at gmail.com>>; Alonso, Juan Carlos <juan.carlos.alonso at intel.com<mailto:juan.carlos.alonso at intel.com>>
Cc: starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>
Subject: Re: [Starlingx-discuss] Deployment Option

Hi Himanshu,
“system host-list” doesn’t “see” your compute node and LLDP won’t work,
because the mgt port on compute node directly connects to mgt port on controller-0 (rather than both connecting to a hub).

Anyway, given you know the MAC of mgt port on compute node, you can have a try to run the following cmd:

# system host-add -n compute-0 -p compute -m <compute_node_mgt_port_mac>

Regards,
yong

From: Himanshu Goyal <himanshugoyal500 at gmail.com<mailto:himanshugoyal500 at gmail.com>>
Date: Monday, 21 January 2019 at 8:14 PM
To: "Alonso, Juan Carlos" <juan.carlos.alonso at intel.com<mailto:juan.carlos.alonso at intel.com>>
Cc: "starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>" <starlingx-discuss at lists.starlingx.io<mailto:starlingx-discuss at lists.starlingx.io>>
Subject: Re: [Starlingx-discuss] Deployment Option

Thanks Juan,

Able to unlock my controller node. But facing Issue in PXE boot of compute node. After unlocking of controller machine not able to see compute host in "system host-list" command.
my controller machine is directly connected to compute machine.

I'm following the below steps
Steps:
1) system host-unlock controller-0
2) system host-list
Output::
[wrsroot at controller-0 ~(keystone_admin)]$ system host-list
+----+--------------+-------------+----------------+-------------+--------------+
| id | hostname     | personality | administrative | operational | availability |
+----+--------------+-------------+----------------+-------------+--------------+
| 1  | controller-0 | controller  | unlocked       | enabled     | available    |
+----+--------------+-------------+----------------+-------------+--------------+

3)  power on my compute machine. And give option to boot from PXE
   my compute machine is directly connected with controller with mgmt port.
  But not able to see host in "system host-list".

4) i tried with system host-add command also, but it is giving below error:
Error:
[wrsroot at controller-0 ~(keystone_admin)]$ system host-add
Host-add Rejected: Cannot add a compute host without specifying a mgmt_ip when static address allocation is configured.

Please suggest me the needful change.

Regards,
Himanshu Goyal


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.starlingx.io/pipermail/starlingx-discuss/attachments/20190123/d8d83be6/attachment-0001.html>