Add templates for Thanos Operator #729
Conversation
Signed-off-by: Saswata Mukherjee <[email protected]>
| - asyncForwardWorkerCount: 50 | ||
| externalLabels: | ||
| replica: $(POD_NAME) | ||
| image: quay.io/thanos/thanos:v0.37.2 |
There was a problem hiding this comment.
I wonder should we template out the image and set this as default. Also we need to use the konflux build here
| kind: ThanosStore | ||
| metadata: | ||
| creationTimestamp: null | ||
| name: default |
There was a problem hiding this comment.
isnt it a case that we want a store per bucket, so in this case two stores and within that three sets of time sharded stores? so in fact in the end i believe we would have a reference to 6 different ThanosStore CRs? Here we appear to have two. One for telemeter and one for mst.
Maybe mst is a better name than default here too but Im not fussy on it
There was a problem hiding this comment.
Also given what I just said I wonder if we should move the resources into individual templates just so its easier to track diffs?
| lazyDownloadStrategy: lazy | ||
| lazyReaderIdleTimeout: 5m | ||
| logFormat: logfmt | ||
| logLevel: info |
There was a problem hiding this comment.
probably can remove these values that will be sensibly defaulted to tidy things up. Again Im not too fussy on it
| shardReplicas: 2 | ||
| shards: 3 | ||
| type: block | ||
| storageSize: 50GiB |
There was a problem hiding this comment.
I think this is way too large. We can go to 2GiB here based on what I currently see in stage and we can even go further lower if we intend (which I believe we do) in time splitting these shards?
| optional: false | ||
| shardingStrategy: | ||
| shardReplicas: 2 | ||
| shards: 3 |
There was a problem hiding this comment.
potentially some misunderstanding, but I thought given what we now know about block sharding we were going to shard by time?
There was a problem hiding this comment.
ah i see the other ThanosStores now at the bottom of the file. In that case this wont make sense though?
| kind: ThanosCompact | ||
| metadata: | ||
| creationTimestamp: null | ||
| name: rhobs |
There was a problem hiding this comment.
we have two identically named compactors here and below!
| fiveMinutes: 365d | ||
| oneHour: 365d | ||
| raw: 365d | ||
| storageSize: "" |
There was a problem hiding this comment.
missing a size here. looks like 2GB will suffice for telemter and 1Gb for mst
| kind: ThanosCompact | ||
| metadata: | ||
| creationTimestamp: null | ||
| name: rhobs |
| matchLabels: | ||
| operator.thanos.io/query-api: "true" | ||
| queryRangeMaxRetries: 3 | ||
| queryRangeSplitInterval: 24h |
There was a problem hiding this comment.
as we discussed previously, I think it makes more sense to go with 2d here
| logFormat: logfmt | ||
| logLevel: info | ||
| name: telemeter | ||
| replicas: 15 |
There was a problem hiding this comment.
overkill for stage, lets go with 6
| key: thanos.yaml | ||
| name: default-objectstorage | ||
| optional: false | ||
| replicas: 6 |
There was a problem hiding this comment.
lets go with 3 as we have now
| imagePullPolicy: IfNotPresent | ||
| logFormat: logfmt | ||
| logLevel: info | ||
| replicas: 6 |
| shardReplicas: 2 | ||
| shards: 3 | ||
| type: block | ||
| storageSize: 100GiB |
Signed-off-by: Saswata Mukherjee <[email protected]>
magefiles/template.go
Outdated
|
|
||
| // Stage images. | ||
| var StageImages = ParamMap[string]{ | ||
| "STORE02W": "quay.io/thanos/thanos:v0.37.2", |
There was a problem hiding this comment.
lets create a const for quay.io/thanos/thanos:v0.37.2 like currentImage or something and point to that in the map. Makes future updates easier
Signed-off-by: Saswata Mukherjee <[email protected]>
Signed-off-by: Saswata Mukherjee <[email protected]>
magefiles/operator.go
Outdated
| Containers: []corev1.Container{ | ||
| { | ||
| Name: "kube-rbac-proxy", | ||
| Image: "gcr.io/kubebuilder/kube-rbac-proxy:v0.16.0", |
There was a problem hiding this comment.
I think we will need to pull an image from RH registry for this.
magefiles/template.go
Outdated
| return v | ||
| } | ||
|
|
||
| const CurrentThanosImageStage = "quay.io/thanos/thanos:v0.37.2" |
There was a problem hiding this comment.
same comment here. I think we can use the konflux fork
| @@ -0,0 +1,341 @@ | |||
| apiVersion: template.openshift.io/v1 | |||
| kind: Template | |||
There was a problem hiding this comment.
we will need a servicemonitor for this. Ideally, because of the way I intend on rolling them out, it should be in its own template.
| valueFrom: | ||
| fieldRef: | ||
| fieldPath: metadata.name | ||
| image: registry.redhat.io/rhosdt/jaeger-agent-rhel8:1.57.0-10 |
There was a problem hiding this comment.
not to worry about this for now but I wonder if we should swap this out for the otel collector at some point
Signed-off-by: Saswata Mukherjee <[email protected]>
Signed-off-by: Saswata Mukherjee <[email protected]>
Signed-off-by: Saswata Mukherjee <[email protected]>
Signed-off-by: Saswata Mukherjee <[email protected]>
No description provided.