These 9 commits are when the Protocol Buffers files have changed:
Commit: | f6646b7 | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
agent, resmgr: nuke agent.GetConfig(). [cherry-picked #986/9c4ccf3] We started propagating the result of the last config update from the resource manager through the agent and the agent probe back to the control plane. Therefore, stop querying the agent for configuration data on the startup path. If we failed to apply the queried configuration, there would be no way to relay that information back to the control plane, unless we added a dedicated interface on the agent side of the gRPC API, which we rather don't do. Since we stopped using agent.GetConfig(), remove it from the protocol, agent and resmgr. Signed-off-by: Krisztian Litkey <krisztian.litkey@intel.com>
The documentation is generated from this commit.
Commit: | f7b62fa | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
pkg/agent: add support for HealthCheck queries. [cherry-picked #986/9f0c4bb] Allow limiting the scope of a HealthCheck query by specifying a query string. Add a constant for querying the status of the last configuration update pushed to the resource manager. When receiving a ConfigStatus query, fetch the result of the last config update pushed to the resource manager and reply according to this. Otherwise keep responding as before. This should allow us to run the original liveness probe unmodified. Signed-off-by: Krisztian Litkey <krisztian.litkey@intel.com>
Commit: | 9c4ccf3 | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
agent, resmgr: nuke agent.GetConfig(). We started propagating the result of the last config update from the resource manager through the agent and the agent probe back to the control plane. Therefore, stop querying the agent for configuration data on the startup path. If we failed to apply the queried configuration, there would be no way to relay that information back to the control plane, unless we added a dedicated interface on the agent side of the gRPC API, which we rather don't do. Since we stopped using agent.GetConfig(), remove it from the protocol, agent and resmgr. Signed-off-by: Krisztian Litkey <krisztian.litkey@intel.com>
The documentation is generated from this commit.
Commit: | 9f0c4bb | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
pkg/agent: add support for HealthCheck queries. Allow limiting the scope of a HealthCheck query by specifying a query string. Add a constant for querying the status of the last configuration update pushed to the resource manager. When receiving a ConfigStatus query, fetch the result of the last config update pushed to the resource manager and reply according to this. Otherwise keep responding as before. This should allow us to run the original liveness probe unmodified. Signed-off-by: Krisztian Litkey <krisztian.litkey@intel.com>
Commit: | ad259a7 | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
agent,config: update for recent protoc/go plugin. Update agent and config servers to compile against bindings generated with more recent protoc and protoc-gen-co. Tested with v3.20.1 and v1.28.0 respective.
Commit: | b51e2e4 | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
resource-manager: update external adjustments, trigger rebalancing. Update and store external adjustments received from the agent. Trigger rebalancing if the cache tells us the update touched some of the containers.
Commit: | 451a230 | |
---|---|---|
Author: | Markus Lehtonen | |
Committer: | Markus Lehtonen |
agent: extend gRPC API with a healthcheck
Commit: | 676716d | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
agent, resource-manager: add error field to config update reply message. Return configuration update errors as a field in the reply message. This allows the agent to differentiate between errors in delivery vs. errors in applying an update. For delivery errors the agent will keep retrying indefinitely, sending an update request at periodic intervals. For errors in applying an update the agent will give up immediately and only try sending a new update when the configration is updated the next time.
Commit: | 06ad1a9 | |
---|---|---|
Author: | Krisztian Litkey | |
Committer: | Krisztian Litkey |
repo: initial commit of code, minimal documentation.