Bug1277038 - Differ between `oc env` updates(removes/adds) the resource(s) envvars and when it doesnt #6104

jhadvig · 2015-11-26T16:00:19Z

When adding/removing envvars with oc env cmd, differ between states when the command actually updates the resource(s) and when it doesnt.
Eg: When updating resource by adding the envvar with a value that is already present in the resource, or removing a envvar that is not present, the printed output will look accordingly:
replicationcontroller "database-1" nothing to update
instead of
replicationcontroller "database-1" updated

added a TC

Wasnt sure if we also wanna print due to which envar the resource wasnt updated (eg1. removing envvar thats not present in the resource || updating envar with the same value), since we would have to revrite some parts of the env.go and will end up with dup. messages in case user would want to update all rcs/dc with wrong envars(eg1).

@fabianofranz thoughts ?

fabianofranz · 2015-11-27T14:37:48Z

[test]

liggitt · 2015-11-27T14:41:21Z

Will conflict/overlap with #5819
How do we treat deletions of resources which don't exist?

jhadvig · 2015-11-27T15:03:14Z

@liggitt I've already spoken with @Kargakis about the overlap and this PR doesn't solve this issue.
So we could either:

merge this and rebase oc: Make env use PATCH instead of PUT #5819
wait until oc: Make env use PATCH instead of PUT #5819 is merged and rebase this PR

Don't have any strong feelings about both solutions. Thoughts ?

0xmichalis · 2015-11-28T02:21:54Z

#5819 supersedes this. The success message there needs fixing though, @jhadvig may want to rebase your changes regarding it once it gets merged.

jhadvig · 2015-11-30T09:17:21Z

@Kargakis yes, once the our PR is merged, will rebase. Thx

jhadvig · 2015-12-02T16:39:47Z

@fabianofranz I've updated the PR after #5819 got merged
Now when deleting envs that aren't present in the pod's container the oc will return:

$ oc env dc/frontend TEST1- TEST2-
environment variable "TEST1" not found in ruby-helloworld container
environment variable "TEST2" not found in ruby-helloworld container
deploymentconfig "frontend" nothing to update

PTAL

liggitt · 2015-12-02T16:49:40Z

What do the label/annotate commands do in this case?

jhadvig · 2015-12-02T17:12:25Z

@liggitt I've already proposed this kind of change for the label cmd to upstream and it was merged aswell kubernetes/kubernetes#17946

Regarding the annotate cmd, its not returning this kind of info(for now)

liggitt · 2015-12-02T17:37:09Z

pkg/cmd/cli/cmd/env.go

+				for _, env := range remove {
+					_, ok := findEnv(c.Env, env)
+					if !ok {
+						infosMsgs[i] = append(infosMsgs[i], fmt.Sprintf("environment variable %q not found in %s container\n", env, c.Name))


So if something has multiple containers, one of which has an envvar, and you remove the envvar, you'll always get a message saying the other container didn't have that var? That doesn't seem necessary.

if user is updating the envars in a pod, by default it will update all the containers. For that reason user should specify --containers flag I guess.

I wouldn't print warnings that an envvar they wanted to remove didn't exist in all containers. If it existed in at least one, that's probably fine

Hmm, how about printing them with glog at some higher verbosity ?

nah, if I ask to remove envvar B, and it exists in one container but not another, I don't consider that failure or warning-worthy

jhadvig · 2015-12-03T10:52:59Z

@liggitt updated
ptal

0xmichalis · 2015-12-03T11:21:48Z

pkg/cmd/cli/cmd/env.go

-			continue
+		outputMsg := "updated"
+		if !updated[i] {
+			outputMsg = "nothing to update"


Instead of doing this and nesting all the code, do this:

if !updated[i] { cmdutil.PrintSuccess(mapper, shortOutput, out, info.Mapping.Resource, info.Name, "didn't need to update") continue } // rest of the code...

Also it would be nice if this (no need for update) was handled by CreateTwoWayMergePatch by returning an empty byte slice and not by us but that's not the case: kubernetes/kubernetes#17022

Actually @liggitt if we pin the size of an empty patch (a couple of bytes) to a constant and check for it, wouldn't it be better?

The size of empty patch 2

It's {}, right? Not sure how I feel about checking that. I think I would rather have updateEnv additionally return a boolean indicating whether any updates were made

jhadvig · 2015-12-04T11:50:34Z

@liggitt @Kargakis I've updated the PR. updateEnv will additionally return bool whether container envvars had been updated
ptal

liggitt · 2015-12-04T15:57:16Z

pkg/cmd/cli/cmd/env.go

@@ -371,7 +384,11 @@ func updateEnv(existing []kapi.EnvVar, env []kapi.EnvVar, remove []string) []kap
 		covered.Insert(e.Name)
 		out = append(out, e)
 	}
-	return out
+	envUpdated := true


instead of using reflection, I'd rather initialize envUpdated to false at the top, and set it to true in the loop where we know we're making changes

Then the line which would be setting envUpdate to false would need to be basically in all three if statements and also at the end of the for loops, due to the logic the out variable is build. Would rather prevent prevent that.

I still would rather set it explicitly... I think some of the complexity would be reduced by handling each case (removed, modified, added) in its own loop

ok, will rewrite the logic and set it explicitly :)

…vars and when it doesnt

jhadvig · 2015-12-08T00:31:57Z

@liggitt cleaned the logic a bit as suggested.
ptal

openshift-bot · 2015-12-08T00:35:23Z

Evaluated for origin test up to 2cbfd4a

openshift-bot · 2015-12-08T01:15:25Z

continuous-integration/openshift-jenkins/test FAILURE (https://ci.openshift.redhat.com/jenkins/job/test_pull_requests_origin/7677/)

liggitt · 2015-12-08T02:21:25Z

pkg/cmd/cli/cmd/env.go

+	var envUpdated bool
+	var updated = make([]bool, len(infos))
+	for i, info := range infos {
+		updated[i] = false


defaults to false already

liggitt · 2015-12-08T02:26:17Z

pkg/cmd/cli/cmd/env.go

-			covered.Insert(e.Name)
-			out = append(out, newer)
-			continue
+			envUpdated = true


only if the value has changed, right?

I guess this is probably a tight enough check

danmcp · 2016-01-20T00:21:00Z

@jhadvig Bump

openshift-bot · 2016-02-12T23:15:23Z

Origin Action Required: Pull request cannot be automatically merged, please rebase your branch from latest HEAD and push again

danmcp · 2016-06-14T13:06:24Z

Closing due to age and inactivity. Feel free to reopen if you are going to continue working on it.

jhadvig force-pushed the BZ_1277038 branch from c7bc4bd to 7c05dd1 Compare November 27, 2015 15:13

openshift-bot added the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 2, 2015

jhadvig force-pushed the BZ_1277038 branch from 7c05dd1 to d712d3c Compare December 2, 2015 16:33

openshift-bot removed the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 2, 2015

jhadvig force-pushed the BZ_1277038 branch from d712d3c to 6693da0 Compare December 2, 2015 17:03

liggitt reviewed Dec 2, 2015
View reviewed changes

jhadvig force-pushed the BZ_1277038 branch from 6693da0 to 90eb66c Compare December 3, 2015 10:52

0xmichalis reviewed Dec 3, 2015
View reviewed changes

jhadvig force-pushed the BZ_1277038 branch from 90eb66c to 9675680 Compare December 4, 2015 11:46

jhadvig force-pushed the BZ_1277038 branch from 9675680 to df78605 Compare December 4, 2015 13:38

liggitt reviewed Dec 4, 2015
View reviewed changes

Bug1277038 - Differ between updates(removes/adds) the resource(s) env…

2cbfd4a

…vars and when it doesnt

jhadvig force-pushed the BZ_1277038 branch from df78605 to 2cbfd4a Compare December 8, 2015 00:30

liggitt reviewed Dec 8, 2015
View reviewed changes

openshift-bot added the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Feb 12, 2016

openshift-bot added needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. and removed needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. labels Apr 30, 2016

openshift-bot added needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. and removed needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. labels May 24, 2016

openshift-bot added needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. and removed needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. labels Jun 7, 2016

danmcp closed this Jun 14, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Bug1277038 - Differ between `oc env` updates(removes/adds) the resource(s) envvars and when it doesnt #6104

Bug1277038 - Differ between `oc env` updates(removes/adds) the resource(s) envvars and when it doesnt #6104

jhadvig commented Nov 26, 2015

fabianofranz commented Nov 27, 2015

liggitt commented Nov 27, 2015

jhadvig commented Nov 27, 2015

0xmichalis commented Nov 28, 2015

jhadvig commented Nov 30, 2015

jhadvig commented Dec 2, 2015

liggitt commented Dec 2, 2015

jhadvig commented Dec 2, 2015

liggitt Dec 2, 2015

jhadvig Dec 2, 2015

liggitt Dec 2, 2015

jhadvig Dec 2, 2015

liggitt Dec 2, 2015

jhadvig commented Dec 3, 2015

0xmichalis Dec 3, 2015

0xmichalis Dec 3, 2015

0xmichalis Dec 3, 2015

jhadvig Dec 3, 2015

liggitt Dec 3, 2015

jhadvig commented Dec 4, 2015

liggitt Dec 4, 2015

jhadvig Dec 6, 2015

liggitt Dec 7, 2015

jhadvig Dec 7, 2015

jhadvig commented Dec 8, 2015

openshift-bot commented Dec 8, 2015

openshift-bot commented Dec 8, 2015

liggitt Dec 8, 2015

liggitt Dec 8, 2015

liggitt Dec 8, 2015

danmcp commented Jan 20, 2016

openshift-bot commented Feb 12, 2016

danmcp commented Jun 14, 2016

Bug1277038 - Differ between oc env updates(removes/adds) the resource(s) envvars and when it doesnt #6104

Bug1277038 - Differ between oc env updates(removes/adds) the resource(s) envvars and when it doesnt #6104

Conversation

jhadvig commented Nov 26, 2015

fabianofranz commented Nov 27, 2015

liggitt commented Nov 27, 2015

jhadvig commented Nov 27, 2015

0xmichalis commented Nov 28, 2015

jhadvig commented Nov 30, 2015

jhadvig commented Dec 2, 2015

liggitt commented Dec 2, 2015

jhadvig commented Dec 2, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jhadvig commented Dec 3, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jhadvig commented Dec 4, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jhadvig commented Dec 8, 2015

openshift-bot commented Dec 8, 2015

openshift-bot commented Dec 8, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

danmcp commented Jan 20, 2016

openshift-bot commented Feb 12, 2016

danmcp commented Jun 14, 2016

Bug1277038 - Differ between `oc env` updates(removes/adds) the resource(s) envvars and when it doesnt #6104

Bug1277038 - Differ between `oc env` updates(removes/adds) the resource(s) envvars and when it doesnt #6104