This post is more than 5 years old
88 Posts
0
220884
2 Nodes of NLB Cluster will detected as 1 Server
I use OME 2.1.0.2635.
Since two days I have a problem with the discovery and inventory of two servers which are an NLB cluster. OME detects the MAC of the NLB as MAC of the servers and interpret this so that there is one server where the name changed. See attachment.
It is not possible to mange both servers with OME.
The Servers are 2008 R2 Std. with OMSA 8.2.0. The NLB has dedicated NICs so I don't understand how an why OME detect the NLB instead the servers itself. I configure the discovery range with the IP addresses of the servers so there should be no hint for OME that there is more than this.
This configuration function several month. The update to OMSA 8.2 was made before some weeks.
Greetings Barbara
MdmBarbarella
88 Posts
0
January 13th, 2016 00:00
Hi Pupul,
today I found the answer for this problem but it is a little bit strange.
At first: this both servers are a high critical system. I avoid it to reboot one of them outside the official maintenance hours if I can.
Yesterday in the afternoon I detected that the server adserv196 has strange errors while I tried to install the new diag software. I decided that a reboot was the only clean way to clear this. I have done this today in the morning.
What then happens was following: The server could not start because it was not able to initialize the iDRAC. Shut off and on of the server don't helped. I must pull the power cables to make the complete machine powerless for 10 sec to reset the iDRAC. After this the server started without errors. The iDRAC is online again.
And the error with the NLB-MAC disappears. Now OME can discover and initialize both servers again.
It seems that the hanging iDRAC causes the problem. I had hanging iDRACs before but never seen such problem. And why this became visible only after the discovery after loading a new catalog stays a secret.
Thank you for your effort. I hope this can help somebody with a similar problem.
Greetings Barbara
DELL-Pupul M
1K Posts
0
January 7th, 2016 05:00
Hi Barbara,
Few initial questions to narrow down the problem:
MdmBarbarella
88 Posts
0
January 7th, 2016 06:00
Hi,
1. ICMP and WMI are enabled.
2. Yes.
3. No. Both servers have different IP addresses and different names (this would be jolly if not...)
4. No. The update to OMSA 8.2 was last year in the end of November. Instead it started instantly after download of the newest catalog. The download of the catalog completed Jan 5, 11:45. The failure I showed in the screenshot appears the first time Jan 5, 11:52. It seems that the inventory after loading the new catalog causes the problem.
The strange thing is, I have another system with NLB configuration where I have no problems with. The difference is the hardware. The servers with this error are R420. The servers what are ok are R610. The systems are comparable because the R420 are the productive systems of a software, the R610 are the test system of the same software. Both systems have the same configuration.
Greetings Barbara
MdmBarbarella
88 Posts
0
January 12th, 2016 01:00
I switched back to OMSA 8.1.0.1 on both servers but the problem persists. It seems that OME causes the phenomenon. But I don't understand why this function over month. There were no changes on OME since a September 2015.
DELL-Pupul M
1K Posts
0
January 12th, 2016 21:00
Hi Barbara,
Are you able to see the service tag of these servers are discovery?
Did you try nslookup or ping -a on the IP addresses? Just want to be sure that there is not stale data in DNS cache. You can clear the DNS by using "ipconfig /flushdns" command if that turns out to the problem.
Can you also try deleting the IP range from OME and perform a new discovery of the individual IP addresses one after another to be sure, make note to check if the service tag is getting populated after discovery of each device?
DELL-Pupul M
1K Posts
0
January 13th, 2016 22:00
Hi Barbara,
Thanks for updating the post. It looks like somehow the MAC address which was present in both the servers was same. How! It is difficult to answer without really looking at the systems. But i would say for now, great that it is solved.
MdmBarbarella
88 Posts
0
January 14th, 2016 00:00
Microsoft NLB-Service creates a virtual MAC address for the NLB cluster. This MAC address is naturally the same on both servers. This virtual MAC address should be only reachable over the second NIC of the servers.
In the OME every server is as its own range with dedicated IP address configured. In this case NIC1.
But the OME detected the virtual MAC of the NLB cluster as MAC address of both servers.
The question is, why. And what has the iDRAC to with this. But I think this will stay a secret.