alanbover / deathnode

89%

master: 85%

LAST BUILD ON BRANCH recover_untracked_terminating_instances

branch: recover_untracked_terminating_instances

CHANGE BRANCH

x

Reset

• recover_untracked_terminating_instances
• add_lifecycle_timeout_reset
• code_refactor
• enable_travis_and_improve_doc
• filter_protected_instances
• filterframeworkcontraint
• fix_crash_when_no_instance_found
• fix_refresh_lifecyclehook_issue
• improve_code_reports
• make_deathnode_tag_configurable
• master
• move_tests_to_convey
• revert-32-code_refactor
• some_refactor
• tasklabels
• use_lifecycles

Recover untracked terminating instances

We have now enabled lifecyclehooks in AWS. This means that if an instance
dies because it goes to unhealthy state (and the autoscaling kills it),
this instance has also go through the termination lifecycle. As deathnode
is not tracking it, the instance would still be alive until the heardbeat
times out, spending unnecesary money and messing with the number of desired
instances.

With this PR, deathnode is able to recover from this situation marking
those instances with DEATHNODE mark