build: Add make nuke; speed-up make start
#5583
Conversation
Signed-off-by: Alex Collins <alex_collins@intuit.com>
Codecov Report
@@ Coverage Diff @@
## master #5583 +/- ##
==========================================
+ Coverage 47.01% 47.10% +0.08%
==========================================
Files 240 240
Lines 15002 15002
==========================================
+ Hits 7053 7066 +13
+ Misses 7048 7038 -10
+ Partials 901 898 -3
Continue to review full report at Codecov.
|
| @@ -444,7 +460,7 @@ $(GOPATH)/bin/goreman: | |||
|
|
|||
| .PHONY: start | |||
| ifeq ($(RUN_MODE),local) | |||
| start: install executor-image controller cli $(GOPATH)/bin/goreman | |||
| start: install controller cli $(GOPATH)/bin/goreman | |||
There was a problem hiding this comment.
this change means that we don't wait for the executor to be built before starting up, useful for UI only work
| controller: PNS_PRIVILEGED=true DEFAULT_REQUEUE_TIME=${DEFAULT_REQUEUE_TIME} LEADER_ELECTION_IDENTITY=local ALWAYS_OFFLOAD_NODE_STATUS=${ALWAYS_OFFLOAD_NODE_STATUS} OFFLOAD_NODE_STATUS_TTL=30s WORKFLOW_GC_PERIOD=30s UPPERIO_DB_DEBUG=${UPPERIO_DB_DEBUG} ARCHIVED_WORKFLOW_GC_PERIOD=30s ./dist/workflow-controller --executor-image ${IMAGE_NAMESPACE}/argoexec:${VERSION} --namespaced=${NAMESPACED} --namespace ${NAMESPACE} --loglevel ${LOG_LEVEL} | ||
| argo-server: UPPERIO_DB_DEBUG=${UPPERIO_DB_DEBUG} ./dist/argo --loglevel ${LOG_LEVEL} server --namespaced=${NAMESPACED} --namespace ${NAMESPACE} --auth-mode ${AUTH_MODE} --secure=$SECURE --x-frame-options=SAMEORIGIN | ||
| ui: yarn --cwd ui install && yarn --cwd ui start | ||
| executor-image : make executor-image |
There was a problem hiding this comment.
we build it here instead
There was a problem hiding this comment.
Do we have to wait to start workflows in this case? Is there a way to know when we are ready to do so? I guess Pods will just take longer to start up, although they could BackOff
alexec
changed the title
make nuke; speed up make startmake nuke; speed-up make start
| @@ -57,7 +55,9 @@ endif | |||
| # * `local` run the workflow–controller and argo-server as single replicas on the local machine (default) | |||
| # * `kubernetes` run the workflow-controller and argo-server on the Kubernetes cluster | |||
| RUN_MODE := local | |||
| K3D := $(shell if [[ "`which kubectl`" != '' ]] && [[ "`kubectl config current-context`" == "k3d-"* ]]; then echo true; else echo false; fi) | |||
| KUBECTX := $(shell [[ "`which kubectl`" != '' ]] && kubectl config current-context || echo none) | |||
There was a problem hiding this comment.
you only get make nuke for KUBECTX=docker-desktop OR KUBECTX=k3d-*
|
|
| @@ -416,6 +420,18 @@ ifeq ($(RUN_MODE),kubernetes) | |||
| kubectl -n $(KUBE_NAMESPACE) scale deploy/argo-server --replicas 1 | |||
| endif | |||
|
|
|||
| # nuke is like "clean" but attempts to return you to a state as if it had never been installed at all | |||
| # this only available in "safe" local development | |||
| ifeq (true,$(filter true, $(DOCKER_DESKTOP) $(K3D))) | |||
There was a problem hiding this comment.
k3d is not guaranteed to be a dev env, but if you're using make nuke you better know what you're doing.
There was a problem hiding this comment.
I hope that "nuke" help people be careful - but k3d should only be desktop - k3s might not
| controller: PNS_PRIVILEGED=true DEFAULT_REQUEUE_TIME=${DEFAULT_REQUEUE_TIME} LEADER_ELECTION_IDENTITY=local ALWAYS_OFFLOAD_NODE_STATUS=${ALWAYS_OFFLOAD_NODE_STATUS} OFFLOAD_NODE_STATUS_TTL=30s WORKFLOW_GC_PERIOD=30s UPPERIO_DB_DEBUG=${UPPERIO_DB_DEBUG} ARCHIVED_WORKFLOW_GC_PERIOD=30s ./dist/workflow-controller --executor-image ${IMAGE_NAMESPACE}/argoexec:${VERSION} --namespaced=${NAMESPACED} --namespace ${NAMESPACE} --loglevel ${LOG_LEVEL} | ||
| argo-server: UPPERIO_DB_DEBUG=${UPPERIO_DB_DEBUG} ./dist/argo --loglevel ${LOG_LEVEL} server --namespaced=${NAMESPACED} --namespace ${NAMESPACE} --auth-mode ${AUTH_MODE} --secure=$SECURE --x-frame-options=SAMEORIGIN | ||
| ui: yarn --cwd ui install && yarn --cwd ui start | ||
| executor-image : make executor-image |
There was a problem hiding this comment.
Do we have to wait to start workflows in this case? Is there a way to know when we are ready to do so? I guess Pods will just take longer to start up, although they could BackOff
Yes. This will always be shorter than current. |
No description provided.