SY480 gen9 are added as monitored to Oneview Composer

[Ten_Admin]

Hello everyone,

Last week a HPE engineer and me did a synergy grow from one to two frames. We ran into some issues but in the end we succeeded. The frame which we already had has only sy480 gen10 compute modules. Before we left the datacenter I moved ten sy480 gen9 compute modules from another frame to the frame we added with the synergy grow. I only checked if all the light on the front were green and called it a day.

Next day I check oneview and I want to add profiles to the gen9's and I find the following. All the gen9's retained the iLO ip (ipv4) they got from the address range of the frame they were taken out of. Some were reporting issues that IML had some errors but they were able to communicate with oneview. Also I noticed that the gen9's were added with status monitored. So I am unable to put a profile on these servers.

I was able to change the ip of the iLO through the console-view option where you can take over the console of a booting compute module and system utillities become available. I gave the iLO an ip (ipv4) from the address range of the frame it is in now. The iLO comes available but the ip is not removed from the address range, so there is a risk the ip will be used again. On one of the other gen9's I tried to do a factory reset of the iLO hoping that after doing a power fuse the compute module would get an ip from oneview address range. But unfortunately this did not happen. Power fuse also does not help getting the compute module added in a managed state. As this is oneview composer this cannot be a license issue.

Firmware is 2020.07.01 for both computes and shared infrastructure and oneview composer is running on 5.20.01. Anyone else had these issues? Google seems to think I am the only one as I do not find any information about this behaviour. 

As I have some time off from work my wife does not allow me to work so a case will have to wait untill I am back at work, but browsing forums and asking questions is allowed

Kind regards,

Raymond

[DanCernese]

Hopefully you can enjoy your time off.  Definitely please open a case when you are back.  This is not typical behavior and ceratinly not common enough to find googling around.  Manually poking IPv4 addresses is a temporary workaround and yes it will eventually cause conflict until the Composer can assign them properly.

After you did a factory reset of the iLO, the composer should act as if the compute was removed and added (although not as thoroughly as an efuse action) and the compute should go through a typical "Adding" cycle-- what error occurs for that add?  I am guessing the add didn't complete, perhaps saying something about not being able to communicate to the iLO?

[Ten_Admin]

Dan,

that is the odd part, there is no error, they are added succesfully but with status monitored. Oneview is also able to communicate with the compute modules. I gues through the ipv6 ip. I can power on, power off, reset iLO. I cannot go to the ilo page itself through the gui and I cannot open a console as that should connect to the ipv4 ip which is not good.

Only way to get to the ilo is ssh to the composer and use the console-view command. Reboot the server and use "esc-F9" and change the ip manually. I also tried a factory reset of the iLO hoping after a refresh or efuse it would get an address from the address range but also did not happen.

I will open a case when I get back to work.

[ChrisLynch · Accepted solution]

The IP address assigned by any external source, or within an LE, is not used for managing a compute node within a Synergy management ring.  Can you verify the frame/enclosure these nodes are in is in a configured Logical Enclosure?  If not, then compute nodes will show up in inventory state when the frame/enclosure is not configured.

Otherwise, a support case and an appliance support dump will be needed to identify why this is happening in your environment.

[DanCernese]

Chris has a good point-- there's a chance your "grow" operation had not completed successfully.

[Ten_Admin]

It was indeed the case that the grow operation was not completed. The whole enclosure was in a monitored state. After we did a reapply configuration the frame was changed into configured state and all the compute nodes were also changed from monitored to managed and given an ip from the address pool and we are able to assign profiles.

Thank you for pointing me in the right direction.