K8s: developments for HW running

[wanyunSu]

Test (linked issue #582 )

issues found:

• controller server correct binding [CS] (K8s: fix logs, root-controller binding and namespace creation #725 )
• thread pinning fails - Interim fix: mimic SSH in k8s [CS] (k8s: thread pinning #617 )
• ~/.drunc.json path resolution [CS] (K8s: fix logs, root-controller binding and namespace creation #725 )
• USER env var setting [CS] (K8s: fix logs, root-controller binding and namespace creation #725 )
• unify k8s json files [MR] ([K8s] Mounting changes, labels, shutdown #679)
• fix PM after cluster upgrade [MR] (K8s Process manager update (after cluster changes) #657)
• move mounts config to json [MR] ([K8s] Mounting changes, labels, shutdown #679)
• mount for df pods [MR] ([K8s] Mounting changes, labels, shutdown #679)
• mount /log for all [MR] ([K8s] Mounting changes, labels, shutdown #679)
• labels for controllers (others?) ([K8s] Mounting changes, labels, shutdown #679)
• rework shutdown procedure [MR] ([K8s] Mounting changes, labels, shutdown #679)
• remnant sleep and magic number

to-do

• test the hardware
• test running a session that writes to network-mounted disk
• compare data files (ssh / k8s) to confirm k8s contains all expected data

Less important

• verify parent_death_pact works in k8s [CS] (k8s : better terminate #622)
• if segment controller dies: respawn and connect to children (error recovery is longer term development) (partially fixed in Add reconnection for children after restart with IP changes #588 )
• drunc-process-manager-shell: should inform once booting sequence is complete
• localhost override when running local session
• better termination process [CS] (k8s : better terminate #622 )

[wanyunSu]

The issues listed above are addressed except for thread pinning which requires further investigation.