Set default redundancy policy to zero

[objectiser]

Fixes #497

Signed-off-by: Gary Brown [email protected]

[objectiser]

@jpkrohling Decided to make it the default regardless of node count - makes it simpler.

@jkandasa Could you test the PR please to make sure fixes the issue.

[jpkrohling]

It would probably be wise to create a ticket to come up with smarter defaults based on the number of nodes, but for now, LGTM.

[jkandasa]

@objectiser can you provide operator image with this PR?

[objectiser]

@jkandasa Will do

[codecov]

Codecov Report

Merging #501 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #501      +/-   ##
==========================================
+ Coverage   91.94%   91.95%   +<.01%     
==========================================
  Files          65       65              
  Lines        3290     3292       +2     
==========================================
+ Hits         3025     3027       +2     
  Misses        185      185              
  Partials       80       80

Impacted Files	Coverage Δ
pkg/strategy/controller.go	97.72% <100%> (+0.03%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 01bb012...e5c9667. Read the comment docs.

[codecov]

Codecov Report

Merging #501 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #501      +/-   ##
==========================================
+ Coverage   91.94%   91.95%   +<.01%     
==========================================
  Files          65       65              
  Lines        3290     3292       +2     
==========================================
+ Hits         3025     3027       +2     
  Misses        185      185              
  Partials       80       80

Impacted Files	Coverage Δ
pkg/strategy/controller.go	97.72% <100%> (+0.03%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 01bb012...e5c9667. Read the comment docs.

[jkandasa]

@objectiser I tested this fix, works as expected,
If we do not pass any redundancyPolicy, jaeger-operator supplies as ZeroRedundancy.
I hope ZeroRedundancy will work with any nodeCount.
I do not see any issue on elasticsearch. It is up and running.

But I see jaeger services are not coming up, maybe another issue?

time="2019-07-04T13:46:54Z" level=error msg="failed to apply the changes" error="CronJob.batch \"simple-prod-es-index-cleaner\" is invalid: spec.jobTemplate.spec.ttlSecondsAfterFinished: Forbidden: disabled by feature-gate" execution="2019-07-04 13:46:54.301508841 +0000 UTC" instance=simple-prod namespace=jkandasa

[objectiser]

@jkandasa Thanks - that other issue is related to #494

[pavolloffay]

This string option is used to specify replica shards. Instead of exposing an integer ECL decided to expose a human readable string. The number of nodes is taken into the account when calculating the replica shards.

jaeger-operator/pkg/storage/elasticsearch.go

Lines 200 to 212 in 0d6bd5f

	func calculateReplicaShards(policyType esv1.RedundancyPolicyType, dataNodes int) int {
	switch policyType {
	case esv1.FullRedundancy:
	return dataNodes - 1
	case esv1.MultipleRedundancy:
	return (dataNodes - 1) / 2
	case esv1.SingleRedundancy:
	return 1
	case esv1.ZeroRedundancy:
	return 0
	default:
	return 1
	}

This should be defaulting to SingleRedundancy to keep backwards compatibility.