)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11090,"name":"Sergii Golovatiuk","email":"sgolovat@redhat.com","username":"holser"},"change_message_id":"97adc693697f7ab9deeaaf9ec461c5fd1d1d639e","unresolved":false,"context_lines":[{"line_number":9,"context_line":"With pcs 0.10 and RHEL/CentOS 8, every node that needs to be"},{"line_number":10,"context_line":"added into a running cluster muste be authenticated first in pcsd."},{"line_number":11,"context_line":""},{"line_number":12,"context_line":"So when replacing a cluster node or automatically scaling up the"},{"line_number":13,"context_line":"cluster, make sure that we run a \"pcs host auth\" command prior to"},{"line_number":14,"context_line":"any \"pcs cluster node add\"."},{"line_number":15,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"7faddb67_55bb5674","line":12,"updated":"2019-08-07 11:28:21.000000000","message":"If/when we replace the node I guess we need to run \"pcs host deauth\" first","commit_id":"8b30131ef70007a050f9f1aafd7cb695bae7a72a"},{"author":{"_account_id":20778,"name":"Damien Ciabrini","email":"dciabrin@redhat.com","username":"dciabrin"},"change_message_id":"18c84200be3c83424aef9fe9da49ddd348b39985","unresolved":false,"context_lines":[{"line_number":9,"context_line":"With pcs 0.10 and RHEL/CentOS 8, every node that needs to be"},{"line_number":10,"context_line":"added into a running cluster muste be authenticated first in pcsd."},{"line_number":11,"context_line":""},{"line_number":12,"context_line":"So when replacing a cluster node or automatically scaling up the"},{"line_number":13,"context_line":"cluster, make sure that we run a \"pcs host auth\" command prior to"},{"line_number":14,"context_line":"any \"pcs cluster node add\"."},{"line_number":15,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"7faddb67_40a7120d","line":12,"in_reply_to":"7faddb67_55bb5674","updated":"2019-08-07 11:47:03.000000000","message":"That\u0027s correct. Actually that step is part of the replacement procedure currently. We do 1. pcs cluster node remove, 2. pcs host deauth.\n\nWe never automated the removal of a node during an stack deploy. So I thin kwe\u0027re not introducing any regression here.","commit_id":"8b30131ef70007a050f9f1aafd7cb695bae7a72a"}],"manifests/corosync.pp":[{"author":{"_account_id":11090,"name":"Sergii Golovatiuk","email":"sgolovat@redhat.com","username":"holser"},"change_message_id":"074d89bdd7c2c565aa97ddf6fc475d7c9aeb189b","unresolved":false,"context_lines":[{"line_number":251,"context_line":"        $node_name \u003d split($node_to_add, \u0027 \u0027)[0]"},{"line_number":252,"context_line":"        if $::pacemaker::pcs_010 {"},{"line_number":253,"context_line":"          exec {\"Authenticating new cluster node: ${node_to_add}\":"},{"line_number":254,"context_line":"            command   \u003d\u003e \"${::pacemaker::pcs_bin} host auth ${node_name} -u hacluster -p ${::pacemaker::hacluster_pwd}\","},{"line_number":255,"context_line":"            timeout   \u003d\u003e $cluster_start_timeout,"},{"line_number":256,"context_line":"            tries     \u003d\u003e $cluster_start_tries,"},{"line_number":257,"context_line":"            try_sleep \u003d\u003e $cluster_start_try_sleep,"}],"source_content_type":"text/x-puppet","patch_set":2,"id":"7faddb67_159cfe0c","line":254,"updated":"2019-08-07 11:27:26.000000000","message":"Interesting how to make this command idempotent.","commit_id":"8b30131ef70007a050f9f1aafd7cb695bae7a72a"},{"author":{"_account_id":20778,"name":"Damien Ciabrini","email":"dciabrin@redhat.com","username":"dciabrin"},"change_message_id":"18c84200be3c83424aef9fe9da49ddd348b39985","unresolved":false,"context_lines":[{"line_number":251,"context_line":"        $node_name \u003d split($node_to_add, \u0027 \u0027)[0]"},{"line_number":252,"context_line":"        if $::pacemaker::pcs_010 {"},{"line_number":253,"context_line":"          exec {\"Authenticating new cluster node: ${node_to_add}\":"},{"line_number":254,"context_line":"            command   \u003d\u003e \"${::pacemaker::pcs_bin} host auth ${node_name} -u hacluster -p ${::pacemaker::hacluster_pwd}\","},{"line_number":255,"context_line":"            timeout   \u003d\u003e $cluster_start_timeout,"},{"line_number":256,"context_line":"            tries     \u003d\u003e $cluster_start_tries,"},{"line_number":257,"context_line":"            try_sleep \u003d\u003e $cluster_start_try_sleep,"}],"source_content_type":"text/x-puppet","patch_set":2,"id":"7faddb67_206d36cc","line":254,"in_reply_to":"7faddb67_159cfe0c","updated":"2019-08-07 11:47:03.000000000","message":"I don\u0027t have a super effective way of not running that command it\u0027s not necessary. Because in order to check whether this is necessary, I would have to check on all the nodes, which is not possible.\n\nHowever I think there are two things which make it fairly idempotent as far as state is concerned.\n\n1. that code only runs when puppet finds out that new nodes must be added. I.e a subsequent deploy would not retrigger it.\n\n2. If for any reason (e.g. puppet breaks right after running this resource) that resource must be re-executed, pcs host auth still succeeds. So there\u0027s no error thrown out.","commit_id":"8b30131ef70007a050f9f1aafd7cb695bae7a72a"}]}