An excellent discussion , Tx for sharing it . What will be the state of apps and argocd , when REDIS get restarted ? , assuming there is no HA configured and its an Autosync enabled environment
It might be a basic question as I am a newbie. where do we set all these tuning parameters in ha/install? 1. Should we configure them via values.yaml and helm install argocd? 2. Should we directly edit the configmaps like argocd-cm? I really want to practice GitOps for ArgoCD install itself where these tuning parameters are captured in Git and can be applied any number of times.. considering that what is a better approach?
If you install argo using helm chart, you can override them using values file. If installed manually, kustomization or want to override them in cluster, argocd-cm and argocd-cmd-params-cm configmaps.
Nice to see more focus on argocd scaling but some of the tunnings presented are inaccurate or just plain wrong like the one about the memory usage of plain manifests with the directory plugin.
I don't know what you mean but based on the docs when using raw manifests with directory plugin, memory usage can be grown up to 300x of the on the rest size.
So you completely recommend against raw manifests? I see that in other talks, this pattern is recommended, as it reduces the compute load on the repo server and also helps show actual changes, rather than abstract changes to Helm and Kustomise values.
Proud of You Saeed
We really need the in-depth topics like this, thankssss 💯
An excellent discussion , Tx for sharing it . What will be the state of apps and argocd , when REDIS get restarted ? , assuming there is no HA configured and its an Autosync enabled environment
It might be a basic question as I am a newbie.
where do we set all these tuning parameters in ha/install?
1. Should we configure them via values.yaml and helm install argocd?
2. Should we directly edit the configmaps like argocd-cm?
I really want to practice GitOps for ArgoCD install itself where these tuning parameters are captured in Git and can be applied any number of times.. considering that what is a better approach?
If you install argo using helm chart, you can override them using values file. If installed manually, kustomization or want to override them in cluster, argocd-cm and argocd-cmd-params-cm configmaps.
Thanks @ssbostan. This is one of the best argocd videos I have watched on RUclips. 👏
I don't know if I'm distracting from the agenda but few seconds in and this kubedemy stuff sounds hella interesting 2 me 🥳🥳
Nice deep dive for scaling ArgoCD. Do you have the repo of all the configs that you showed in the last slide to understand the moving parts ?
Nice to see more focus on argocd scaling but some of the tunnings presented are inaccurate or just plain wrong like the one about the memory usage of plain manifests with the directory plugin.
I don't know what you mean but based on the docs when using raw manifests with directory plugin, memory usage can be grown up to 300x of the on the rest size.
Brilliant 👏 👏
You are perfect!
So you completely recommend against raw manifests? I see that in other talks, this pattern is recommended, as it reduces the compute load on the repo server and also helps show actual changes, rather than abstract changes to Helm and Kustomise values.
I preferred rendered manifest model or rendered helm with kustomization. Check CNCF talks for more info about these patterns.
Legit! Thanks!