How to forcibly reprovision an Oracle Private Cloud(VM Compute) Appliance (PCA/OVCA) Compute Node

Asset ID:	1-71-1636246.1
Update Date:	2017-07-09
Keywords:

Solution Type Technical Instruction Sure

Solution 1636246.1 : How to forcibly reprovision an Oracle Private Cloud(VM Compute) Appliance (PCA/OVCA) Compute Node

Applies to:

Private Cloud Appliance - Version 1.0.1 to 2.1.1 [Release 1.0 to 2.0]
Oracle VM - Version 3.2.4 to 3.2.10 [Release OVM32]
Oracle Virtual Compute Appliance X3-2 Hardware - Version All Versions to All Versions [Release All Releases]
Oracle Virtual Compute Appliance X4-2 Hardware - Version All Versions to All Versions [Release All Releases]
Linux x86-64

Goal

This article describes how to forcibly reprovision a problematic Oracle Private Clould(Virtual Compute) Appliance (PCA/OVCA) Compute Node (CN).

Solution

PCA/OVCA Manager continually monitors the PCA/OVCA node database and displays the current status of Compute Nodes (CNs) via the PCA/OVCA Management Dashboard.
Whilst a Compute Node (CN) remains in the running state, the PCA/OVCA Dashboard Reprovision option is unavailable (greyed out).
Under some circumstances, PCA/OVCA Support may deem it necessary for a Compute Node to be reprovisioned.
To be able to re-activate the PCA/OVCA Dashboard Reprovision option for a Compute Node, the status of ilom-<CN node name> must be changed to a non-running state within the PCA/OVCA node database.

Steps to facilitate Reprovisioning of an PCA/OVCA Compute Node

**NOTE** 2.1.x and later have provided a force reprovision function in pca-admin command:

# pca-admin reprovision --force compute-node <Computer Node>

1. Login (ssh, putty, etc.) to the current primary PCA/OVCA Management Node (MN) using the MN Virtual IP (VIP)

2. Verify the details of the problematic Compute Node e.g. name, status

Prior to version 2.0.5

# /usr/sbin/ovca-node-db list type=compute

or command pca-node-db for version 2.0.5.

# /usr/sbin/pca-node-db list type=compute

mac=00:10:xx:xx:xx:xx ip=192.168.4.6 state=DEAD         link=00:10:xx:xx:xx:xx name=ovcacn11r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.10 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn08r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.12 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn07r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.11 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn14r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.7 state=INITIALIZING link=00:10:xx:xx:xx:xx name=None       type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.13 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn09r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.9 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn10r1 type=compute
mac=00:10:xx:xx:xx:xx ip=192.168.4.8 state=RUNNING      link=00:10:xx:xx:xx:xx name=ovcacn12r1 type=compute

In this example, Compute Node ovcacn14r1 shall be reprovisioned.

3. Update the status of the problematic Compute Node and it's iLOM SP in the PCA/OVCA node database

Prior to version 2.0.5

# /usr/sbin/ovca-node-db update name=ilom-ovcacn14r1 to state=dead # *** lowercase dead
# /usr/sbin/ovca-node-db update name=ovcacn14r1 to state=DEAD # *** UPPERCASE DEAD

or command pca-node-db for version 2.0.5

# /usr/sbin/pca-node-db update name=ilom-ovcacn14r1 to state=dead # *** lowercase dead
# /usr/sbin/pca-node-db update name=ovcacn14r1 to state=DEAD # *** UPPERCASE DEAD

NODE: 00:10:xx:xx:xx:xx
    * name = ilom-ovcacn14r1
    * ovmm_physical_disks = [('0004fb0000180000be1923e169775e38', '3600605b005b655501aaa700507d7ea02')]
    * ip = 192.168.4.11
    * ovmm_iscsi = ['iqn.1988-12.com.oracle:87f64b4177a4']
    * state = dead (WAS: running)
    * ovmm_id = e0:2e:77:aa:00:10:ff:ff:ff:ff:ff:ff:ff:20:00:07
    * link = 00:10:e0:2e:77:aa
    * ovmm_ethernet = [('0004fb0000200000b1e2843adbe5f79a', 'eth0 on ovcacn14r1'), ('0004fb00002000001bdb66fb54114ad8', 'eth1 on ovcacn14r1'), ('0004fb00002000008e6c1e576a0e1097', 'eth2 on ovcacn14r1'), ('0004fb000020000032db99089a26a12c', 'eth3 on ovcacn14r1'), ('0004fb000020000052dd4a1966ae9e57', 'eth4 on ovcacn14r1'), ('0004fb00002000005652b8312cb803df', 'eth4B on ovcacn14r1'), ('0004fb0000200000f9d010c114d80bcb', 'eth5 on ovcacn14r1'), ('0004fb00002000009743ed54604fcdea', 'eth5B on ovcacn14r1'), ('0004fb0000200000470b5ee65241bcb5', 'eth6 on ovcacn14r1'), ('0004fb0000200000660a6f4b91425f00', 'eth6B on ovcacn14r1'), ('0004fb000020000017d6543d7c0226fb', 'eth7 on ovcacn14r1'), ('0004fb00002000000672a7dd6d412637', 'eth7B on ovcacn14r1')]
    * ovmm_name = ovcacn14r1
    * type = ilom
    * ovmm_bondport = [('0004fb000020000010894ceed1a372ed', 'bond0 on ovcacn14r1'), ('0004fb0000200000927a41738d88e338', 'bond2 on ovcacn14r1')]

4. Login to the PCA/OVCA Dashboard to reprovision the Compute Node

Login to the PCA/OVCA Dashboard - https://<MN VIP>:7002/ovca
Refresh the PCA/OVCA Dashboard after which the Reprovision option will be available for the relevant Compute Node.
Proceed to reprovision the affected Compute Node accordingly.

Note that the IP address of the iLOM SP and corresponding Compute Node may change once reprovisioned.

References

<NOTE:1676039.1> - [PCA] Provioning Nodes on Oracle PCA 1.0.2 Fails due to Xsigo "ioprofile" being Out of Sync and iSCSI Initiator Name has been Truncated
<NOTE:1591197.1> - How to determine OVCA or PCA Management Node, Compute Node, Switch and ILOM service processor IP addresses

Attachments

This solution has no attachment