Health Manager details update
Updated the Health Manager details with corrections and format. Change-Id: I6b69096c890cadde59ceaf924a5854572387b1c7
This commit is contained in:
parent
17c2bf4e9e
commit
1d5364aed2
@ -33,7 +33,7 @@ Kuryr Controller might get to a broken state due to problems like:
|
||||
unable to connect with services it depends on and they being not healthy.
|
||||
|
||||
It is important to check health of these services so that Kubernetes and
|
||||
its users know when Kuryr Controller it is ready to perform its networking
|
||||
its users know when Kuryr Controller is ready to perform its networking
|
||||
tasks. Also, it is necessary to check the health state of Kuryr components in
|
||||
order to assure Kuryr Controller service is alive. To provide these
|
||||
functionalities, Controller's Health Manager will verify and serve the health
|
||||
@ -41,7 +41,7 @@ state of these services and components to the probes.
|
||||
|
||||
Besides these problems on the Controller, Kuryr CNI daemon also might get to a
|
||||
broken state as a result of its components being not healthy and necessary
|
||||
configurations not present. It is essencial that CNI components health and
|
||||
configurations not present. It is essential that CNI components health and
|
||||
configurations are properly verified to assure CNI daemon is in a good shape.
|
||||
On this way, the CNI Health Manager will check and serve the health state to
|
||||
Kubernetes readiness and liveness probes.
|
||||
@ -54,7 +54,7 @@ and talk to Neutron, since these are services needed by Kuryr Controller.
|
||||
These checks will assure the Controller readiness. The other endpoint, will
|
||||
verify the health state of Kuryr components and guarantee Controller liveness.
|
||||
|
||||
The CNI Health Manager also provides two endpoins to Kubernetes probes.
|
||||
The CNI Health Manager also provides two endpoints to Kubernetes probes.
|
||||
The endpoint that provides readiness state to the probe checks connection
|
||||
to Kubernetes API and presence of NET_ADMIN capabilities. The other endpoint,
|
||||
which provides liveness, validates whether IPDB is in working order, maximum
|
||||
@ -62,17 +62,17 @@ CNI ADD failure is reached, health of CNI components and existence of memory
|
||||
leak.
|
||||
|
||||
.. note::
|
||||
The CNI Health Manager will be started with the check for memory leak disabled.
|
||||
In order to enable, set the following option in kuryr.conf to a limit value
|
||||
of memory in MiBs.
|
||||
The CNI Health Manager will be started with the check for memory leak
|
||||
disabled. In order to enable, set the following option in kuryr.conf to a
|
||||
limit value of memory in MiBs.
|
||||
|
||||
[cni_health_server]
|
||||
max_memory_usage = -1
|
||||
|
||||
The CNI Health Manager is added as a process to CNI daemon and communicates
|
||||
to the other two processes i.e. Watcher and Server with a shared boolean object,
|
||||
which indicates the current health state of each component.
|
||||
to the other two processes i.e. Watcher and Server with a shared boolean
|
||||
object, which indicates the current health state of each component.
|
||||
|
||||
The idea behind these two Managers is to combine all the necessary checks in
|
||||
servers running inside Kuryr Controller and CNI pods to provide the checks result
|
||||
to the probes.
|
||||
servers running inside Kuryr Controller and CNI pods to provide the result of
|
||||
these checks to the probes.
|
||||
|
Loading…
Reference in New Issue
Block a user