Running Argo CD?

Argo CD stuck OutOfSync or waves not rolling? We fix it.

GitOps looks great on the diagram and then your app sits OutOfSync for two days, a ComparisonError cites a cryptic chart line, or an app-of-apps rollout stops halfway through sync wave three. We walk the manifests, the controller logs and the resource status until the cause is visible and sync starts behaving again.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

argocd-app-controller.log

still failing

What it is

Argo CD at a glance

Argo CD is a declarative GitOps controller for Kubernetes. It watches Git repositories for desired state, compares that state against what is running in the cluster, and either reports a diff or syncs the cluster back toward the repository. It lives inside the cluster as a set of controllers with an API server, a UI and a CLI.

The project is a graduated Cloud Native Computing Foundation project under the Apache 2.0 license. Typical uses include multi-cluster application rollouts, the app-of-apps pattern for managing hundreds of applications with one parent manifest, and promotion pipelines where staging and production reflect different Git revisions.

Type
GitOps controller for Kubernetes
First released
2018
License
Apache 2.0
Maintained by
The Argo project, under the CNCF
Typically used for
Declarative Kubernetes application rollout, app-of-apps patterns, multi-cluster delivery

The honest version

Argo CD pros and cons

No technology is all upside. This is where Argo CD earns its place and where it tends to cause trouble.

Where Argo CD is strong

  • Git is the single source of truthReviewable, auditable changes land in a repository, and the controller applies them. You stop wondering what someone kubectl-applied last Friday.
  • Honest diff viewThe UI shows the live-to-desired diff per resource, which turns most drift conversations into a short click through the tree rather than a hunt.
  • App-of-apps scales to large estatesOne parent Application manages many children, which lets a platform team roll out an update across dozens of services without a per-service commit.
  • Supports multiple manifest formatsPlain YAML, Helm, Kustomize, Jsonnet and config-management plugins all slot into the same sync loop, so teams do not have to pick one and stop.
  • Sync waves and hooks for orderingAnnotations control the order resources go in and out, so CRDs install before the controllers, and migrations run before the deploys that need them.
  • Runs where your cluster runsNo external SaaS dependency to roll out to your clusters, and the agent model covers multiple remote clusters from one Argo instance.

Where Argo CD bites

  • ComparisonError messages are terseUpstream chart failures, missing values files and post-renderer crashes all surface as ComparisonError with little context, and the real cause is only visible in repo-server logs.
  • Drift fights between tools are painfulIf anything else (a Terraform module, a Helm release outside Argo, a cluster add-on) mutates the same resource, Argo will revert it on every sync and nobody will be happy.
  • RBAC is its own languageThe argocd-rbac-cm ConfigMap uses a path-style policy syntax that is easy to get wrong, and a misplaced wildcard quietly grants or blocks more than intended.
  • Sync waves do not mean strict dependencyLower waves apply first and Argo waits for health, but complex ordering that assumed strict queue semantics breaks when two resources live in the same wave and race.
  • Repo-server performance limitsRendering hundreds of Helm charts on every refresh pressures CPU and memory on the repo-server pods, which shows up as deadlines exceeded and apps that never finish comparing.
  • Secrets still need another toolArgo does not solve the secrets problem; teams usually pair it with SealedSecrets, External Secrets Operator or a KMS-backed CSI driver, which is one more thing to run and upgrade.

Sound familiar?

Common Argo CD problems we fix

The Argo CD problems that bring people to us, and what you have once each one is fixed.

OutOfSync: Secret/db-creds not found

App goes OutOfSync right after sync

A manifest Argo expects but cannot find in the cluster, usually because the secret is created by a different operator or lives outside the Application's tracked resources.

Ignore-differences or source-of-truth corrected

ComparisonError: duplicate key

A Helm or Kustomize render fails

A values file with repeated keys, a kustomization patch pointing at a missing path, or a template function that returns empty, with the controller unable to produce target state to compare against.

Manifest rendering clean again

cannot apply CRD before controller is ready

App-of-apps fails on the first rollout

A parent Application that installs a CRD-providing chart and immediately applies custom resources that need that CRD, with the second apply failing because the controller is still starting.

Sync waves and health checks ordered correctly

sync-wave 3 waiting on resource health

Sync waves stall mid-rollout

A wave that will not finish because a Deployment never reports healthy, often because of a probe misconfiguration, missing image or a failed pre-sync hook that blocks the rest of the plan.

Wave completing, health honest

context deadline exceeded

Repo-server runs out of time comparing manifests

A monorepo with hundreds of charts or a Kustomize overlay that walks the filesystem too far, with refresh cycles timing out and apps sitting stale.

Repo-server sized and tuned for the workload

RBAC: permission denied

Users cannot sync their own applications

A role mapping in argocd-rbac-cm that does not include the project or namespace a team actually uses, so the UI shows the app but every sync click comes back denied.

RBAC policy aligned with team structure

webhook ignored: signature mismatch

Argo misses Git webhooks

A Git provider webhook configured with the wrong secret or pointing at the wrong path, so Argo polls instead of pushing, and changes land slowly during busy hours.

Webhook wired and verified

What we do

Argo CD help, from rescue to steady rollouts

Ping us for one stuck Application, or hand us the whole GitOps estate and the clusters behind it.

Rescue
An app that will not sync, a parent Application that stalls mid-wave, or a ComparisonError that nobody can decipher. We settle the controller and then explain what broke.
App-of-apps design
Parent and child Application layout, project boundaries and sync waves arranged so a platform team can roll out changes across many services without surprises.
Repo-server tuning
CPU, memory, caching and parallelism on the repo-server pods matched to the real manifest workload, with deadline-exceeded errors gone.
RBAC and multi-team projects
argocd-rbac-cm policies, projects and SSO integration arranged so teams get the access they need and nothing more, with policy tests in CI.
Secrets workflow
SealedSecrets, External Secrets Operator or a KMS-backed CSI driver wired in so secrets stop living in commits or in the wrong hands.
Managed Tech
Sync health watched, upgrades rehearsed on a non-prod instance, and someone to call when a webhook stops firing on a Friday afternoon.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Argo CD alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Argo CD questions

Something else? Call+1 737 727 7327

Why does our app go OutOfSync immediately after syncing?

Usually something outside Argo is mutating the resource, or the manifests render differently between CI and the cluster because of a missing value or a plugin. We identify the field that changes, decide whether Argo should ignore it or whether the other tool should stop touching it, and the drift stops.

What is the right way to use sync waves?

Lower numbers first, with health checks on the resources the next wave will depend on. CRDs usually go in wave -1, controllers in wave 0, custom resources in wave 1, and migrations in a pre-sync hook. Treating waves as hard dependencies for same-wave resources is the common trap.

How should app-of-apps be laid out?

A parent repository with one Application per team or product, each pointing at the team's own repository, inside clearly named Projects. That keeps RBAC simple and lets teams move at their own pace without the parent becoming a bottleneck.

Do we need Argo CD if we already use Helm?

If deploys are done by a human running helm upgrade from a laptop, Argo gives you a review trail, drift detection and consistent rollouts for free. If your Helm releases are already driven from CI into every cluster and audited, the win is smaller and might not justify the operational overhead.

How much does Argo CD help cost?

The quote is at no charge. Describe the repos, the clusters and the sync you are stuck on, and we respond with a number before we touch anything.

Free estimate

Argo CD giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.