Bug 965490

Created attachment 750962 [details]
development.log when rhc domain show

Running the migration script on a single devenv results in a successful 'rhc domain show'

This can be reproduced after migrated on devenv_3258

1. Launch devenv-stage_353
2. Prepare test data
3. Add V1 marker to node
#touch /var/lib/openshift/.settings/v1_cartridge_format
4. scp -r ~/devenv-local and /etc/yum.repos.d/* from devenv_3258 instance to same location on devenv-stage_353 instance
5. yum update -y --enablerepo devenv-local
6. Run oo-admin-clear-pending-ops
7. Remove the v1 marker from the node
#rm -f /var/lib/openshift/.settings/v1_cartridge_format
8. Restart mcollective
#service mcollective restart
9. Clear the broker cache
#oo-admin-broker-cache --clear --console
10. Restart the rhc-broker
#server rhc-broker restart
12. Run migrate-mongo-2.0.28 
13. Run rhc-admin-migrate --version 2.0.28

After migration the acount with one ruby-1.8 app and one jbossas+switchyard app can not communicate with server any more, the error was same with the description.

According to development.log, it was blocked at 'get_profile_for_component' method.

Created attachment 751652 [details]
development.log

Not sure what's going on but have recreated. Doesn't seem to be related to the gear. Seeing nothing in development.log. Can't do anything from rhc.

[root@ip-10-10-42-16 ~]# rhc domain show
The server did not respond correctly. This may be an issue with the server configuration or with your connection to the server
(such as a Web proxy or firewall). Please verify that you can access the OpenShift server https://localhost/broker/rest/api
[root@ip-10-10-42-16 ~]# rhc domain show -d
DEBUG: Using config file /root/.openshift/express.conf
DEBUG: Authenticating with RHC::Auth::Token
DEBUG: Connecting to https://localhost/broker/rest/api
DEBUG: Getting all domains
DEBUG: Client supports API versions 1.1, 1.2, 1.3, 1.4
DEBUG: Using token authentication
DEBUG: Request GET https://localhost/broker/rest/api
DEBUG: SSL Verification failed -- Using self signed cert
DEBUG:    code 500  519 ms
DEBUG: Response did not include a message from server: invalid character at "<!DOCTYPE "
The server did not respond correctly. This may be an issue with the server configuration or with your connection to the server
(such as a Web proxy or firewall). Please verify that you can access the OpenShift server https://localhost/broker/rest/api
[root@ip-10-10-42-16 ~]# service httpd restart
Stopping httpd:                                            [  OK  ]
Starting httpd: [Wed May 22 11:03:38 2013] [warn] module passenger_module is already loaded, skipping
[Wed May 22 11:03:38 2013] [warn] NameVirtualHost *:443 has no VirtualHosts
[Wed May 22 11:03:38 2013] [warn] NameVirtualHost *:80 has no VirtualHosts
                                                           [  OK  ]

The problem here is that the name of the switchyard cartridge changed from V1 -> V2.  I'm making changes so that the name and other information in the manifest will be identical.

PR submitted.

Commit pushed to master at https://github.com/openshift/origin-server

https://github.com/openshift/origin-server/commit/b19157be0fd96e1b1d78a1c92396d776ba0c9304
Fix bug 965490

Retest on migration from devenv_stage_353 to devenv_3290 ---Failed
After migration, the account(swithyard added) could connect server well(rhc domain show)

But when check more, can't control the switchyard cartridge well, the server development.log show;
2013-05-30 05:23:55.901 [DEBUG] DEBUG: server results: Failed to get cartridge 'switchyard-0.6' from  in gear 536256880632503637901312: Cartridge directory not found for switchyard-0.6 (pid:19297)
2013-05-30 05:23:55.901 [ERROR] Node execution failure (invalid exit code from node).  If the problem persists please contact Red Hat support. (pid:19297)

The switchyard cartridge directory deleted after migration actually.

[qgong@dhcp-10-94 openshift-quickstarter]$ rhc cartridge stop -a qjbosseap -c switchyard-0.6
Unable to complete the requested operation due to: Node execution failure (invalid exit code from node).  If the problem persists please contact Red Hat support..
Reference ID: 061e1becdd28d205f0b4e725f81d8446

Multiple PRs submitted, tldr:

1. The V1 switchyard cartridge _does not_ create a directory for switchyard.  It only adds switchyard to the JBoss AS/EAP module path.
2. In V2, there is a discrete directory which will contain but EAP and AS env vars in its env directory.
3. I had to infer the presence of switchyard in an app in a special case for the migration.  Now, switchyard should be accurately detected and both a switchyard directory and a switchyard-0.6 symlink to that directory will be present in the migrated application.

Commit pushed to master at https://github.com/openshift/li

https://github.com/openshift/li/commit/953ee9ba1fcf64f8128adf78a27267797fcec761
Bug 965490: Fix V1 -> V2 migrations for switchyard

Verified on migration from devenv_stage_353 to devenv_3296
After migration, the account connect server well, and could control switchyard well.
[root@dhcp-10-94 smygaeapp]# rhc cartridge stop  -a qjbosseap -c switchyard-0.6
RESULT:
switchyard-0.6 stopped