Creating new vio lpar vhost numbers 2

[phorbiuz]

Ok, we have a 570 and I need to create a new lpar on it via the VIO server.

Using the HMC (v7) I highlighed the server, and from the menu went through Configuration, Create Logical Partition, AIX. It gave this new partition the id of 7. I've given it CPU, memory, network etc etc. Hostname is ukepu111.

In the Logical Partition Profile Properties for the VIO server I've created virtual scsi adapters for the new partition stating adapter id's 70 and 71 connecting to ukepu111(7), connecting adapter is numbered 2 and 3.

I've then gone onto the VIO server and created the new virtual target device logical volume, but when I go to add this to the new lpar there is no ne vhost number for it.

How do I go about getting a new vhost number for the new partition?

Thanks.

 

[p5wizard]

If you have only modified the VIOS's LPAR profile then that VIO server will not know about the new vhost adapters until you shutdown and re-activate it from the HMC.

If you want to go without a VIOS restart, you can redo the same configuration like in the profile under DLPAR menu from the HMC, then you can add the vhost adapters "on the fly".


HTH,

p5wizard

 

[Aixexplore]

Hi Phorbuiz,

If you were not allowed to restart the VIO , Then create Virtual SCSI from DLPAR, than should able see vhost instantly, Once you have vhost#, you can run command
$mkvdev -vdev "lvname or hdiskx" -vadapter vhost# -dev ukepu111


Thanks !

 

[ogniemi]

thx.

Q1: I know the advantage choosing DLPAR method (no VIO shutdown+new_activation is required) but is the any disadvantage comparing this method to VIO shutdown+new_activation? VIO Server or Clients reboot will keep the configuration but shutdown+startup not if the profiles were not adjusted, right?

Q2: If the VIO shutdown+new_activation has to be done, what is the procedure to avoid shutting down clients using VIO resources? (of course with dual VIO configuration)

 

[p5wizard]

A1&2 - No need for VIOS reboot just for this action. Just do the work twice - once in the VIOS's profile and once in DLPAR menu. Make double sure you define the vhost adapters in the profile and in the DLPAR menu exactly the same and then you're all set.

If you do need to shutdown a VIOS and the following conditions are met:

1 - You have two of them.
For Virtual ethernet:
2 - SharedEthernetAdapterFailover configuration (tested and failover works, perhaps adjusted spanning tree configuration on your network hardware i.e. rapid or fast reconfiguration)
For Virtual SCSI
3 - dual paths to the LUNs via the two VIOSs
4 - no SCSI reservation on both VIOSs
5 - attribute hcheck_interval=60 for all VSCSI disks on all VIO clients

Then you just reboot the VIOSs one by one. Allow a couple of minutes between the two reboots and verify that all VSCSI disk paths are back to normal after max 60s following reboot of each VIOS (use [tt]lspath[/tt] command to verify path status). If you have still disabled paths on a VIO client then you may have to activate them manually. run this command for each disabled path:

[tt]chpath -l hdiskX -p vscsiY -s enable[/tt]

This simply means that you have VSCSI disks on a VIO client which still have the default hcheck_interval=0. So run this command for each disk involved:

[tt]chdev -l hdiskX -a hcheck_interval=60 -P[/tt]

The VIO clients involved will then still need a reboot at next maintenance window to activate the changed settings. ([tt]chdev -P[/tt] only changes the ODM, not the active device as that can't be changed when it is in use - e.g. when the VG it is part of is varied on).

You may still want to try out this with all VIO clients down except one test VIO client just to get used to this scenario.


HTH,

p5wizard