-
Notifications
You must be signed in to change notification settings - Fork 6
Crowbar issues and workarounds deails
This page contains more detailed instructions for working around various issues
This is manifested as recipe failures to locate cluster members, and caused by too much data in index inside chef. Two options:
-
use short domain names - the simple case
-
increase the index field size (see below)
To increase max field size (all on the admin node):
-
edit
- on ubuntu: /var/lib/chef/solr/conf/solrconfig.xml
- on redhat: /var/chef/solr/conf/solrconfig.xml
-
find the field maxFieldLength and change the value to 200000
-
repeat for both occurrence
-
restart chef-solr. On mesa 1.2 issue: bluepill chef-server restart chef-solr
-
rebuild the index by issuing the following: knife index rebuild
Update the rabbitmq cookbook files in /opt/dell/chef/cookbooks/rabbitmq/providers/user.rb and /opt/dell/chef/cookbooks/rabbitmq/recipes/default.rb with the changes here:
https://github.com/dellcloudedge/barclamp-crowbar/commit/ab9683f21c4a5addbfbe9ff89ab15fde625f88b7
execute (as root): knife cookbook upload rabbitmq -o /opt/dell/chef/cookbooks/
Update the nova cookbook in /opt/dell/chef/cookbooks/nova/recipes/rabbit.rb with the changes here:
https://github.com/dellcloudedge/barclamp-nova/commit/525ec99952679f3d39861be34485d1d90d6afc8d
execute (as root):
knife cookbook upload nova -o /opt/dell/chef/cookbooks/