[operator] Add documentation for coredump preservation and pdb (#533)

* Create coredump_preservation_guide.md

* Finish updating coredump_preservation_guide.md

* Add guide for Pod Disruption Budget (PDB) Feature

* Polish coredump_preservation_guide.md

* Polish pdb_guide.md

doc/user/coredump_preservation_guide.md

@@ -0,0 +1,41 @@
+## Coredump Preservation
+
+To facilitate debugging in the event of a crash, Nebula Graph automatically preserves coredumps generated by all services (metad, graphd and storaged) when the coredump preservation feature is enabled. The coredumps are stored on a seperate volume generated by the Nebula operator during cluster installation. You are able to specify the size of the generated volume and the amount of time the coredump should be kept before it is automatically deleted. 
+
+### Prerequisites
+* Nebula operator is installed
+* Nebula operator version >= 1.8.2
+
+### Using the Coredump Preservation Feature
+The coredump preservation feature can be used by either installing the Nebula cluster using the provided [helm chart](https://github.com/vesoft-inc/nebula-operator/tree/add-documentation/charts/nebula-cluster) in this repo or by adding the required values directly in your nebulacluster.yaml configuration. The steps for using each method are listed below.
+
+#### 1. Using the Coredump Preservation Feature with the Provided Nebula Cluster Helm Chart:
+The coredump preservation feature is enabled by default when installing with the provided Nebula cluster helm chart. The default values for the size of the generated volume and the amount of
+time the coredump is kept is 5Gi and 72h respectively. To use this feature with the default values simply install Nebula cluster with the provided helm chart following the instructions [here](nebula_cluster_guide.md). To customize the values, look for the `coredumpPreservation:` section under `nebula` in the [values.yaml](https://raw.githubusercontent.com/vesoft-inc/nebula-operator/refs/heads/master/charts/nebula-cluster/values.yaml) file of the helm chart and modify the values below.
+```yaml
+nebula:
+  coredumpPreservation:
+    # Maximum time for the coredump to be kept on the seperate coredump volume
+    # before being automatically deleted.
+    maxTimeKept: 72h
+    # wheather to enable the coredump preservation feature
+    enable: true
+    # Size of the coredump volume
+    volumeSize: 5Gi
+```
+
+#### 2. Using the Coredump Preservation Feature Directly in Your Nebulacluster.yaml Configuration:
+To enable the coredump preservation feature, please add the following parameters to your nebulacluster.yaml file before running `kubectl apply`. The `maxTimeKept` and `storage` parameters can be modified as you see fit, with `maxTimeKept` being at least 1 minute.
+```yaml
+spec:
+  coredumpPreservation:
+    # Maximum time for the coredump to be kept on the seperate coredump volume
+    # before being automatically deleted.
+    maxTimeKept: 72h
+    # Specs for the coredump volume 
+    volumeSpecs:
+      resources:
+        requests:
+          # Size of the coredump volume
+          storage: 5Gi
+```

doc/user/pdb_guide.md

@@ -0,0 +1,91 @@
+## Pod Disruption Budget
+The Nebula Cluster Helm chart has a feature to automatically start pod disruption budgets (PDBs) for all services it controls (metad, graphd, storaged) if requested by you. PDBs help minimize latency issues by ensuring a minimum number of required pods for the requsted service are available at all times during voluntary node outages. You may specify the minimum pods available or the maximum pods unavailable for a given service and whether to apply the PDB to all nebula graph services or just a particular service.
+
+### Prereqisites
+* Nebula operator is installed
+* Nebula operator version >= 1.8.2
+
+### Warning
+The Nebula Cluster pod disruption budget feature is based on the [kubernetes PDB](https://kubernetes.io/docs/tasks/run-application/configure-pdb/), so it will only be enforced for voluntary disruptions (i.e. draining a node for maintance, rolling update). Any involutary distuptions (i.e. node down, forced update) will cause the PDB to be overridden. There's nothing Nebula can do to prevent this due to the limitations of kubernetes.
+
+### Using the Pod Disruption Budget Feature In the Given Helm Chart
+
+#### Enableing the pdb feature:
+The PDBs feature is disabled for all Nebula Cluster services by default. To enable and use this feature for all services, look for the `pdb:` section in the [values.yaml](https://raw.githubusercontent.com/vesoft-inc/nebula-operator/refs/heads/add-documentation/charts/nebula-cluster/values.yaml) file for the given helm chart and set the `enable` field for all three services to `true` like the example below. 
+```yaml
+pdb:
+  graphd:
+    enable: true
+    minAvailable: 2
+    maxUnavailable: -1
+  metad:
+    enable: true
+    minAvailable: 3
+    maxUnavailable: -1
+  storaged:
+    enable: true
+    minAvailable: 3
+    maxUnavailable: -1
+```
+
+To enable this feature for a particular service (i.e. metad), only set the `enable` field for that service to `true` and Leave the other `enable` fields as `false` as shown in the example below. This can be done for any 1 service or a combination of 2 services.
+```yaml
+# The pdb is only enabled for metad. The enable fields for both graphd and storaged are set to false.
+pdb:
+  graphd:
+    enable: false
+    minAvailable: 2
+    maxUnavailable: -1
+  metad:
+    enable: true
+    minAvailable: 3
+    maxUnavailable: -1
+  storaged:
+    enable: false
+    minAvailable: 3
+    maxUnavailable: -1
+```
+
+#### Setting the minAvailable and maxUnavailable Fields:
+##### 1. minAvailable:
+The `minAvailable` field is used to specify that the PDB should enforce a mimimum number of pods required to be running at a given time for a given service. To use the `minAvailable` field for a given service set it to a positive integer and set the `maxUnavailable` field to `-1` as shown in the example below.
+```yaml
+# The minAvailable field is used for all 3 services as they're all set to positive values
+pdb:
+  graphd:
+    enable: true
+    minAvailable: 2
+    maxUnavailable: -1
+  metad:
+    enable: true
+    minAvailable: 3
+    maxUnavailable: -1
+  storaged:
+    enable: true
+    minAvailable: 3
+    maxUnavailable: -1
+```
+
+##### 2. maxUnavailable:
+The `maxUnavailable` field is used to specify that the PDB should enforce a maximum number of pods that can be down at a given time for a given service. To use the `maxUnavailable` field for a given service set it to a positive integer and set the `minAvailable` field to `-1` as shown in the example below.
+```yaml
+# The maxUnavailable field is used for all 3 services as they're all set to positive values
+pdb:
+  graphd:
+    enable: true
+    minAvailable: -1
+    maxUnavailable: 1
+  metad:
+    enable: true
+    minAvailable: -1
+    maxUnavailable: 1
+  storaged:
+    enable: true
+    minAvailable: -1
+    maxUnavailable: 1
+```
+
+##### Notes: 
+1. Only 1 of `minAvailable` or `maxUnavailable` may be set to a positive integer at a time for a given service. Setting both fields to positive integers for a given service will result in `minAvailable` being used.
+2. An error will result if both `minAvailable` and `maxUnavailable` are set to `-1` for a given service.
+3. It's ok to use `minAvailable` for one service and `maxUnavailable` for another.