Start a Conversation

Unsolved

This post is more than 5 years old

609

July 7th, 2008 04:00

ECC - Master Agent Host UUID lenght =0

Using :
ECC 6.0.0.0.357
IBM AIX 5.3

the host was on controlcenter hosts list we wanted to change the hostname so we stoped the process of master agent using
ps -ef | grep agent
then deleted the content of /var/ECC/*
then deleted the host from ECC hosts list
then change the hostname = > server rebooted
after reboot we restart the Master Agent again

but:
Host has not show any more on the ECC hosts list
Host.UID under /var/ECC is empty (0 sizo)
cert folder under /var/ECC/cert is empty no cer files inside

errors in MGA.log


#######################################################################
SDK 13:48:06 C P W 1 (4.94 MB) swrx_connectable_socket::read => Recv failed with 73: Connection
- reset by peer
SDK 13:48:06 SDK 76 P E 1 (4.94 MB) swrx_connectable_socket::peer_address => Call to getpeername
- failed with 76: Socket is not connected
SDK 13:48:06 C P W 1 (4.94 MB) cecDataComm::readData => Error in cecDataComm::readData(13),
- reading from 0.0.0.0:0. Expected to read 13 bytes. Read 0 bytes. Location was Header -
- InitRead. (Network error occurred. ) Check caller's log for more details.
SDK 13:48:06 C I W 1 (4.94 MB) sstkb::init => Error receiving the reply for InitPacket: Error in
- cecDataComm::readData(13), reading from 0.0.0.0:0. Expected to read 13 bytes. Read 0 bytes.
- Location was Header - InitRead. (Network error occurred. ) Check caller's log for more
- details.
SDK 13:48:06 C P T 1 (4.94 MB) lowLevelClose => Closed socket 6. (1 open socket(s))
MGA 13:48:06 MGA 1 I E 1 (4.94 MB) Initializing SST => Unsuccessful call to SST init, automatic
- retry in 15 seconds
SDK 13:48:21 C P T 1 (4.94 MB) OpenSocket => Opening socket 6. (2 open socket(s))
SDK 13:48:21 C I F 1 (4.94 MB) sstkb::init => Local IP address is set to 172.24.82.9 from live
- socket
SDK 13:48:21 C P F 1 (4.94 MB) swrx_system::getHostVendor => Found HostVendor: "IBM". Length=3.
SDK 13:48:21 C P F 1 (4.94 MB) swrx_system::getHostModel => Found HostModel: "IBM,9117-MMA".
- Length=12.
SDK 13:48:21 C P F 1 (4.94 MB) swrx_system::getUUID => Found Host UUID: "". Length=0.
SDK 13:48:21 C I F 1 (4.94 MB) sstkb::init => HostName='t24shcmb'; HostGUID='t24shcmb.ac185209'
SDK 13:48:21 C I F 1 (4.94 MB) sstkb::init => Starting Initialization: Agent:
- t24shcmb.ac185209_MGA: Server Host: 172.16.73.5
SDK 13:48:21 C I F 1 (4.94 MB) sstkb::init => SST is listening on port 5798
SDK 13:48:21 C P W 1 (4.94 MB) cecLevel0::GetCertificateID => Unable to read from certificate id
- file: unable to open file
SDK 13:48:21 C P W 1 (4.94 MB) cecLevel0::GetCertificate => Unable to read from certificate file:
- unable to open file
SDK 13:48:21 C P F 1 (4.94 MB) sstkb::init => No certificate found on this host - will request a
- new certificate from server
SDK 13:48:21 C P F 1 (4.94 MB) cecDataComm::sendAllData. => Send To: '172.16.73.5:5799'


#######################################################################

please note that:
No certificate found on this host - will request a new certificate from server
swrx_system::getUUID => Found Host UUID: "". Length=0.


Any Ideas ....

6 Posts

July 7th, 2008 05:00

Sure you have a good point which is I should delete only FILES not the cert directory ... and I will keep this note to further host name changes: D

But the case now yes I have tried many times to remove and reinstall master agent but no use.....

Is their any way to make ECC server itself forget about this host.... So it can be a real installation from scratch

Or is there a way to manually make the ECC server initiate a new certificate for this server?

2 Intern

 • 

385 Posts

July 7th, 2008 05:00

Not ideal, but have you tried simply removing the ECC agent and reinstalling from scratch? Certainly not graceful, but it almost seems like you deleted something that maybe you should not have deleted (like directories or something along those lines.

In fact I took a peek at an AIX host and noticed that cert is a subdirectory under /var/ECC and according to the Primus case about changing host names you are only supposed to delete the files in /var/ECC.

My guess - you need the cert directory and the files contained within. You could try creating this directory, but might be best just to reinstall and next time only delete the files in /var/ECC.

That darn recursive delete strikes again ;)

55 Posts

July 7th, 2008 08:00

Hi,

System UUID, is set in the system BIOS, and if the hosts showing the UUID as empty, all have it set to all zero in the system BIOS.

I would suggest to check emc167772 knowledgeBase article.

Nimai

2 Intern

 • 

385 Posts

July 7th, 2008 11:00

When you stop the agents and delete the server from the host list it should remove all agent information from ECC. There should not be anything else you need to do unless ECC itself has a problem and is failing to actually do the delete.

Did you install using the host installation tools or just running the generic ECC installation process? If you are doing the former you want to make sure you uninstall using the same process. Otherwise make sure you remove all of the ECC directories after you are done.

Other than that unless that UUID conflict applies per the other reply I'd suggest you open a case with SAC and let them do some deeper digging.

2 Intern

 • 

1.3K Posts

July 10th, 2008 04:00

here is what we follow for HP-UX

1.Stop all agents on the host, including the Master Agent.
2.Delete the affected host(s) from the ControlCenter Console.
3.For Unix hosts, delete all files in the /var/ECC directory.
4. Cleanup /etc/log/* .
5.Check the agent installation directory (Defaults: /usr/ecc/exec for Unix, C:\ECC\exec for Windows) and delete all files with the following extensions:
* .COMFILE
* .uid
* .port
6. Start the Master Agent.
7. Using ControlCenter Console, start all other Agents on the host.
No Events found!

Top