Ravindra BagaleCourses & study guides मराठी

Ravi Sir's Cards › DevOps

Chapter 19: Kubernetes Services, Ingress and Volumes: Cards

मराठी cards · Open the chapter

Revision cards

Why Pods need a stable front door. Pods come and go; their IPs change every time.
Revision card 1 of 12: Why Pods need a stable front door · Download PNG · Open card page

Lesson: 19.1 Why Pods need a stable front door

ClusterIP: load balance inside the cluster. Most services only need to be reached by other services.
Revision card 2 of 12: ClusterIP: load balance inside the cluster · Download PNG · Open card page

Lesson: 19.2 ClusterIP — load balance inside the cluster

NodePort: a port on the node (for a lab). In a lab you need a quick way in from your laptop.
Revision card 3 of 12: NodePort: a port on the node (for a lab) · Download PNG · Open card page

Lesson: 19.3 NodePort — a port on the node, for a lab

LoadBalancer: when a cloud should publish it. Public apps need a real, stable entry from the internet.
Revision card 4 of 12: LoadBalancer: when a cloud should publish it · Download PNG · Open card page

Lesson: 19.4 LoadBalancer — when a cloud should publish it

ExternalName: an alias to a name outside. Apps shouldn't hardcode outside hostnames that may change.
Revision card 5 of 12: ExternalName: an alias to a name outside · Download PNG · Open card page

Lesson: 19.5 ExternalName — an alias to a name outside

Service DNS step by step. Wrong names cause “could not resolve host” errors.
Revision card 6 of 12: Service DNS step by step · Download PNG · Open card page

Lesson: 19.6 Service DNS, step by step

Ingress: a hostname (not a free public domain). One entry point for many web apps saves cost and adds routing.
Revision card 7 of 12: Ingress: a hostname (not a free public domain) · Download PNG · Open card page

Lesson: 19.7 Ingress — a hostname, not a free public domain

Volumes inside a Pod. Container files disappear on restart, and containers in a Pod may need to share files.
Revision card 8 of 12: Volumes inside a Pod · Download PNG · Open card page

Lesson: 19.8 Volumes inside a Pod

PersistentVolume, Claim and StorageClass. Databases need disks that outlive Pods.
Revision card 9 of 12: PersistentVolume, Claim and StorageClass · Download PNG · Open card page

Lesson: 19.9 PersistentVolume, claim and StorageClass

Safety: Services, Ingress and storage. Exposed services and lost disks are common incidents.
Revision card 10 of 12: Safety: Services, Ingress and storage · Download PNG · Open card page

Lesson: 19.10 Safety

From a Git commit to a running Deployment. Seeing the whole path makes each tool's role clear.
Revision card 11 of 12: From a Git commit to a running Deployment · Download PNG · Open card page

Lesson: 19.11 From a Git commit to a running Deployment

Lab: a name, DNS, one claim. One lab ties Services, DNS and storage together.
Revision card 12 of 12: Lab: a name, DNS, one claim · Download PNG · Open card page

Lesson: 19.12 Lab — name, DNS, one claim

Interview cards

Why do you need a Kubernetes Service? Pod IPs are temporary; a Service gives a stable name/IP and load-balances to matching, ready Pods.
Interview card 1 of 12: Why do you need a Kubernetes Service? · Download PNG · Open card page

Lesson: 19.1 Why Pods need a stable front door

What is ClusterIP? The default Service type: an internal virtual IP and DNS name, load-balancing to Pods, not reachable from outside.
Interview card 2 of 12: What is ClusterIP? · Download PNG · Open card page

Lesson: 19.2 ClusterIP — load balance inside the cluster

When would you use NodePort? For labs or simple on-prem setups to reach a Service via any node's IP and a port in 30000-32767.
Interview card 3 of 12: When would you use NodePort? · Download PNG · Open card page

Lesson: 19.3 NodePort — a port on the node, for a lab

LoadBalancer Service vs Ingress? LoadBalancer: one cloud LB per Service (L4). Ingress: one entry routing many hostnames/paths to many Services (HTTP).
Interview card 4 of 12: LoadBalancer Service vs Ingress? · Download PNG · Open card page

Lesson: 19.4 LoadBalancer — when a cloud should publish it

What is an ExternalName Service? A Service that returns a CNAME to an external DNS name; no proxying, no Pods, no ports.
Interview card 5 of 12: What is an ExternalName Service? · Download PNG · Open card page

Lesson: 19.5 ExternalName — an alias to a name outside

How do Pods find Services by name? CoreDNS resolves service.namespace.svc.cluster.local; short names work within the same namespace.
Interview card 6 of 12: How do Pods find Services by name? · Download PNG · Open card page

Lesson: 19.6 Service DNS, step by step

Why isn't my Ingress working? Usually no Ingress controller installed, wrong ingressClassName, DNS not pointing at it, or the backend Service/port is wrong.
Interview card 7 of 12: Why isn't my Ingress working? · Download PNG · Open card page

Lesson: 19.7 Ingress — a hostname, not a free public domain

What is emptyDir? A volume created when a Pod starts and deleted when the Pod goes; shared by all containers in the Pod.
Interview card 8 of 12: What is emptyDir? · Download PNG · Open card page

Lesson: 19.8 Volumes inside a Pod

PV vs PVC vs StorageClass? PVC: a request. PV: the actual storage. StorageClass: the recipe that creates PVs on demand.
Interview card 9 of 12: PV vs PVC vs StorageClass? · Download PNG · Open card page

Lesson: 19.9 PersistentVolume, claim and StorageClass

How do you keep Kubernetes networking and storage safe? Keep internal services ClusterIP, add TLS on Ingress, use NetworkPolicies, and Retain important volumes.
Interview card 10 of 12: How do you keep Kubernetes networking and storage safe? · Download PNG · Open card page

Lesson: 19.10 Safety

Explain the path from commit to running Pod. PR → CI tests → build and push image (SHA tag) → update the Deployment image → Kubernetes rolls out new Pods → health checks.
Interview card 11 of 12: Explain the path from commit to running Pod. · Download PNG · Open card page

Lesson: 19.11 From a Git commit to a running Deployment

Prove your Service DNS and storage work. From a Pod, nslookup the Service; write data, delete the DB Pod, and read the data back from the PVC.
Interview card 12 of 12: Prove your Service DNS and storage work. · Download PNG · Open card page

Lesson: 19.12 Lab — name, DNS, one claim

Comparison cards

ClusterIP vs NodePort. Both are Kubernetes Service types.ClusterIP: an address only inside the cluster.NodePort: also opens a port (30000-32767) on every node.
Comparison card 1 of 3: ClusterIP vs NodePort · Download PNG · Open card page

Lesson: 19.2 ClusterIP — load balance inside the cluster

LoadBalancer vs Ingress. LoadBalancer: a Service type; the cloud creates one load balancer for it.Ingress: HTTP rules (host, path) that route one entry point to many Services.
Comparison card 2 of 3: LoadBalancer vs Ingress · Download PNG · Open card page

Lesson: 19.7 Ingress — a hostname, not a free public domain

emptyDir vs PersistentVolumeClaim. emptyDir: scratch folder that exists only while the Pod exists.PVC: a claim for a real disk that outlives Pods.
Comparison card 3 of 3: emptyDir vs PersistentVolumeClaim · Download PNG · Open card page

Lesson: 19.9 PersistentVolume, claim and StorageClass