Hi ya,
we've migrated our Hyperic server from one machine to another (inherently upgrading it from 4.4.0 to 4.5.1), following the protocol
given in http://communities.vmware.com/message/1918172#1918172 .
This involves re-importing the old server's hqdb into the empty hqdb of the new server, which to the informed reader of the question perhaps explains some problems described below ;-)
All existing plus all newly installed agents on "normal platforms" work happily with the new server, but we have problems with "Network devices" (SNMP data sources). They do show up with all their config details (as configured once in the old server), but we don't get any metric data from them and their availability is red/unknown.
Trying to reconfigure them via "Configure platform" results in the Hyperic UI error
The configuration has not been set for this resource due to : Invalid configuration: Error reported by Agent @ 127.0.0.2:2144: Unauthorized
after pressing "OK", and the server's local agent.log shows
2011-02-08 11:46:45,521 ERROR [Thread-0] [SSLConnectionListener] Rejecting client from /127.0.0.2: Passed an invalid auth token (128...
I've re-installed and several times also re-configured the local agent on the new hyperic server (bin/agent.sh setup), to no avail.
What I can see is that the local agent talks with the server and successfully gets a new auth token each time, but the token shown as refused in the agent.log is always another one, as if the config of my "Network device" still contains a (hidden?) configuration record pointing to the old server's local agent, causing Hyperic server to send this old agent's auth token instead of the new one by which he knows the new server's local agent.
I have then deleted one "Network device" named "kronos" to recreate and configure it from scratch.
Adding a new Network device named "kronos" shows that in fact there is a choice which agent should do its monitoring (field Agent connection:).
And there I have two identical records with 127.0.0.2:2144 (followed by all other "truly" remote agents). These two local entries are indistinguishable, but one could be the "echo" of my old server's local agent, couldn't it?
Chosing either one of the doublette entries and clicking OK yields a huge java error (see Hyp4.5.1-UI-error_addingNetworkDevice.txt). Chosing the (now remote) agent of the old server machine yields the UI error
This name is already in use.
... also pointing to the agent somehow still knowing "I've had a platform named kronos", doesn't it?
Unfortunately, reconfiguring a "Network device" does not offer at all that Agent connection: field for changing it to the new local agent, so I don't have a chance to get to this "hidden" configuration record as to which agent should be used.
Anyhow, I need to get the "Network devices" back to work - do you happen to have any ideas?
TIA - cheers
we've migrated our Hyperic server from one machine to another (inherently upgrading it from 4.4.0 to 4.5.1), following the protocol
given in http://communities.vmware.com/message/1918172#1918172 .
This involves re-importing the old server's hqdb into the empty hqdb of the new server, which to the informed reader of the question perhaps explains some problems described below ;-)
All existing plus all newly installed agents on "normal platforms" work happily with the new server, but we have problems with "Network devices" (SNMP data sources). They do show up with all their config details (as configured once in the old server), but we don't get any metric data from them and their availability is red/unknown.
Trying to reconfigure them via "Configure platform" results in the Hyperic UI error
The configuration has not been set for this resource due to : Invalid configuration: Error reported by Agent @ 127.0.0.2:2144: Unauthorized
after pressing "OK", and the server's local agent.log shows
2011-02-08 11:46:45,521 ERROR [Thread-0] [SSLConnectionListener] Rejecting client from /127.0.0.2: Passed an invalid auth token (128...
I've re-installed and several times also re-configured the local agent on the new hyperic server (bin/agent.sh setup), to no avail.
What I can see is that the local agent talks with the server and successfully gets a new auth token each time, but the token shown as refused in the agent.log is always another one, as if the config of my "Network device" still contains a (hidden?) configuration record pointing to the old server's local agent, causing Hyperic server to send this old agent's auth token instead of the new one by which he knows the new server's local agent.
I have then deleted one "Network device" named "kronos" to recreate and configure it from scratch.
Adding a new Network device named "kronos" shows that in fact there is a choice which agent should do its monitoring (field Agent connection:).
And there I have two identical records with 127.0.0.2:2144 (followed by all other "truly" remote agents). These two local entries are indistinguishable, but one could be the "echo" of my old server's local agent, couldn't it?
Chosing either one of the doublette entries and clicking OK yields a huge java error (see Hyp4.5.1-UI-error_addingNetworkDevice.txt). Chosing the (now remote) agent of the old server machine yields the UI error
This name is already in use.
... also pointing to the agent somehow still knowing "I've had a platform named kronos", doesn't it?
Unfortunately, reconfiguring a "Network device" does not offer at all that Agent connection: field for changing it to the new local agent, so I don't have a chance to get to this "hidden" configuration record as to which agent should be used.
Anyhow, I need to get the "Network devices" back to work - do you happen to have any ideas?
TIA - cheers