[
https://issues.apache.org/jira/browse/VCL-16?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14303635#comment-14303635
]
ASF subversion and git services commented on VCL-16:
----------------------------------------------------
Commit 1656840 from [~arkurth] in branch 'vcl/trunk'
[ https://svn.apache.org/r1656840 ]
VCL-16
Added utils.pm::get_reservation_management_node_hostname. This is used to
display more useful information when a cluster request fails.
Changed utils.pm::set_reservation_lastcheck to allow multiple reservation IDs
to be passed. Updated call in VIM_SSH.pm.
Updated utils.pm::reservation_being_processed to also retrieve the parent
reservation ID from the database for cluster requests. Duplicate processes were
being forked if a child reservation already ran before the parent finished.
Removed call to update_request_state in State.pm::reservation_failed. This gets
handled by state_exit. It was not checking if the reservation was the parent.
Updated State.pm::wait_for_all_reservations_to_begin to display information
about which management nodes the reservations belong to.
Updated State.pm::wait_for_child_reservations_to_exit to check for both a
computerloadlog 'begin' and 'exited' entry for each reservation. It was only
checking for 'exited'. As a result, the loop ran until it timed out if any
child reservation process had never started and added a 'begin' entry.
VCL-767
Updated utils.pm::determine_remote_connection_target to check the argument for
a '@' sign. run_scp_command is passing user@x.x.x.x. This was failing. If it
detects this, the 'user@' is removed from the argument.
> Enhance cluster reservation process
> -----------------------------------
>
> Key: VCL-16
> URL: https://issues.apache.org/jira/browse/VCL-16
> Project: VCL
> Issue Type: Improvement
> Components: vcld (backend), web gui (frontend)
> Reporter: Josh Thompson
> Assignee: Andy Kurth
> Fix For: 2.4
>
>
> the cluster reservation process is somewhat buggy in both the frontend and
> the backend - this needs to be cleaned up to work solidly
> -- log entry is not getting an ending state
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
