Description of problem: rhq storage not discovered / not auto-imported by server Version-Release number of selected component (if applicable): jon 3.1.2 upgraded using rhqctl to build f4a3f5e How reproducible: Steps to Reproduce: 1. install jon 3.1.2 (postgres) 2. upgrade jon 3.1.2 to rhq 4.8 using rhqctl upgrade --from-agent-dir agentDir --from-server-dir Jon3.1.2ServerDir 3. rhqctl stop 4. rhqctl start --storage 5. migrate data from jon 3.1.2 to Cassandra Actual results: Storage node not auto-discovered -> http://d.pr/i/el1h And cassandra deamon is visible in discovery queue -> http://d.pr/i/oCQ7 Expected results: storage node discovered and auto-inventoried Additional info:
Assuming this is a regression it must be recent as I am not seeing anything like this locally...
Yes, this is a regression - was not visible for previous builds.
Before stopping in reproduction step 3, are you sure the agent was connected and completed its discovery?
I am not sure. Please move it to ON_QA if a retest is needed.
I have not been able to reproduce this behavior, so I moving back to ON_QA to be retested.
Hi Jay, Please get the screen-shots of agent before migration (http://d.pr/i/3qgw) with rhq storage pugin screen-shot (http://d.pr/i/I7sQ). Env details will paste privately. Regards, Armine H.
I successfully upgraded from RHQ 4.5.1 to RHQ 4.8.0-SNAPSHOT. After the upgrade the Storage node was discovered and auto-imported, and the alert template was injected and applied. This was on Windows. There were some bug fixes applied to get my upgrade working successfully although none had to do with the actual auto-import. I did solve two issues that were non-windows related. First, I added the --storage-config option to the upgrade command; this allows the upgrade from pre-48 to specify initial non-default values for storage props (which I needed). Second, I solved a problem with the agent upgrade if the agent was in the default location (no --from-agent-dir being specified). If you are still seeing bad behavior perhaps we can go through the steps together. Let me know.
As an aside, the fixes I added are all in master as of: commit 72f114cb01f3286696be68912280906b1954febe Author: Jay Shaughnessy <jshaughn> Date: Tue Jun 18 15:59:02 2013 -0400 Fix upgrade issues - fix agent upgrade of existing agent in the default location (i.e. no --from-agent-dir specified) - On Windows, ensure exit codes are passed back from the storage installer and rhqctl scripts. The former is very important as otherwise errors in the storage installer do not prevent an attempt to upgrade/install the server.
Tried for Linux upgrade fom 4.5 to 4.8 - the same issue is present. please get screen-shot attached.
Created attachment 762883 [details] storage_node_autoImport
upgrade from 4.5.1 to 48 worked fine -> http://d.pr/i/UAmm. Previous upgrade was on 4.5snapshot which didn't work. Conclution made - upgrade only from rhq-> rhq and jon->jon and not use snapshot versions. Closing bug as wontfix since the scenarios issue appears are not supported.