Update loki distributed for v1.32 #181
Conversation
lukeskywlkr
changed the title
| # admin: admin | ||
| type: s3 | ||
| s3: | ||
| s3: http://minio:[email protected]:9000 |
There was a problem hiding this comment.
Is there a way to specify a secret for the credentials to be used for s3 on this new chart? Maybe we could use the same secret we create with the logging-operated package (or create a dedicated one, to maintain this package as atomic as possible) mounted as env somewhere.
I know we are currently doing as committed here in the current release, I'm just hoping we could change it
There was a problem hiding this comment.
I think it is a good idea, I'm exploring this possibility since, according to the documentation, it seems possible: https://grafana.com/docs/loki/latest/configure/#use-environment-variables-in-the-configuration
We can inject the secret contents as environment variables and then reference them from the Loki config.
There was a problem hiding this comment.
It would be great to have it in a secret. Having said that, pointing to the internal minio could work as a default value but it is important that we can override it later from the distribution, like we do now:
https://docs.kubernetesfury.com/docs/reference/OnPremises#specdistributionmoduleslogginglokiexternalendpoint
| # S3 configuration (when type is "s3") | ||
| s3: | ||
| # S3 endpoint URL | ||
| endpoint: http://minio:[email protected]:9000 |
| ``` | ||
|
|
||
| With the `loki-stack-built.yaml` file, check differences with the current `deploy.yml` file and change accordingly. | ||
| With the `loki-distributed-built.yaml` file and `loki-built.yaml` file, check differences with the current `deploy.yml` file and change accordingly. |
There was a problem hiding this comment.
Does this mean that the two outputs of the helm template commands should be identical? I'm not sure I get why we should template twice with the two different charts
There was a problem hiding this comment.
The two outputs from the helm command won't be the same. We need to take resources that are not created with the loki-distributed chart but are needed for the new loki version (e.g. the index-gateway). Since the version 3.0 Grafana has implemented new Loki components that need to be deployed using the distributed mode. At the same time some components are no more generated by the helm-loki chart (e.g. ServiceMonitors). Hence we need to cherry pick the resources generated. For future work I think we can do a refactor to completely align with the new chart, moving the ServiceMonitors to a custom manifest to avoid to lose them.
There was a problem hiding this comment.
let's add this comment to the maintenance guide.
I'm not a fan of mixing the two charts. I'd rather use the new one and add via kustomize what is missing, like we do in the other modules.
Is it too much effort to do this now @lukeskywlkr?
There was a problem hiding this comment.
I'm working on it, I'll be back with more info ASAP
| Starting with the vTBD of the Logging Core Module Loki version has been bumped to 3.4.2. Please refer to [`loki documentation`](https://grafana.com/docs/loki/v3.4.x/setup/upgrade/) | ||
| for the complete release notes. |
There was a problem hiding this comment.
This is nice to know, but it does not tell me anything about breaking changes. Is this the right place or should it be under the "New features" section?
| ``` | ||
|
|
||
| With the `loki-stack-built.yaml` file, check differences with the current `deploy.yml` file and change accordingly. | ||
| With the `loki-distributed-built.yaml` file and `loki-built.yaml` file, check differences with the current `deploy.yml` file and change accordingly. |
There was a problem hiding this comment.
let's add this comment to the maintenance guide.
I'm not a fan of mixing the two charts. I'd rather use the new one and add via kustomize what is missing, like we do in the other modules.
Is it too much effort to do this now @lukeskywlkr?
| # admin: admin | ||
| type: s3 | ||
| s3: | ||
| s3: http://minio:[email protected]:9000 |
There was a problem hiding this comment.
It would be great to have it in a secret. Having said that, pointing to the internal minio could work as a default value but it is important that we can override it later from the distribution, like we do now:
https://docs.kubernetesfury.com/docs/reference/OnPremises#specdistributionmoduleslogginglokiexternalendpoint
| persistence: | ||
| # -- Enable creating PVCs which is required when using boltdb-shipper | ||
| enabled: true |
There was a problem hiding this comment.
do we need to set this to true? we are moving away from boltdb aren't we?
There was a problem hiding this comment.
we need to remember to align the configuration file also on the distribution side https://github.com/sighupio/fury-distribution/blob/main/templates/distribution/manifests/logging/patches/loki-config.yaml.tpl
And don't forget to test that the build output is the same with kustomize v5.6.0 as it was with v3.10.0 |
Summary 💡
Update
loki-distributedto the latest loki version (3.4.2)Closes: https://github.com/sighupio/product-management/issues/567
Description 📝
loki-distributedchart to0.80.2helm-lokichart version6.28.0loki-distributedchart as base (using image tag3.4.2)to introduce new components from thehelm-lokichartloki-distributednomenclaturemax_label_names_per_seriesto 30 since the new default is 15Breaking Changes 💔
shared_store,shared_store_key_prefix,max_transfer_retriesTests performed 🧪
Future work 🔧
Complete the chart migration, if possible, in order to keep the alignment with the mainstream development.