From e2810559efd5d403cca8362112ea3c998de46769 Mon Sep 17 00:00:00 2001
From: Viktor Barzin <vbarzin@gmail.com>
Date: Sat, 8 Jun 2024 19:22:35 +0000
Subject: [PATCH] add ollama deployment with a ui [ci skip]

---
 modules/kubernetes/main.tf            |   5 +
 modules/kubernetes/ollama/main.tf     | 163 ++++++++++++++++++++++++++
 modules/kubernetes/ollama/values.yaml |  18 +++
 3 files changed, 186 insertions(+)
 create mode 100644 modules/kubernetes/ollama/main.tf
 create mode 100644 modules/kubernetes/ollama/values.yaml

diff --git a/modules/kubernetes/main.tf b/modules/kubernetes/main.tf
index 68b360a0..3f0dcb22 100644
--- a/modules/kubernetes/main.tf
+++ b/modules/kubernetes/main.tf
@@ -455,3 +455,8 @@ module "servarr" {
 #   source = "./dnscat2"
 #   # tls_secret_name = var.tls_secret_name
 # }
+
+module "ollama" {
+  source          = "./ollama"
+  tls_secret_name = var.tls_secret_name
+}
diff --git a/modules/kubernetes/ollama/main.tf b/modules/kubernetes/ollama/main.tf
new file mode 100644
index 00000000..4feb2356
--- /dev/null
+++ b/modules/kubernetes/ollama/main.tf
@@ -0,0 +1,163 @@
+variable "tls_secret_name" {}
+
+resource "kubernetes_namespace" "ollama" {
+  metadata {
+    name = "ollama"
+  }
+}
+
+module "tls_secret" {
+  source          = "../setup_tls_secret"
+  namespace       = "ollama"
+  tls_secret_name = var.tls_secret_name
+}
+resource "kubernetes_persistent_volume_claim" "ollama-pvc" {
+  metadata {
+    name      = "ollama-pvc"
+    namespace = "ollama"
+  }
+
+  spec {
+    access_modes = ["ReadWriteOnce"]
+    resources {
+      requests = {
+        storage = "30Gi"
+      }
+    }
+    volume_name = "ollama-pv"
+  }
+}
+
+resource "kubernetes_persistent_volume" "ollama-pv" {
+  metadata {
+    name = "ollama-pv"
+  }
+  spec {
+    capacity = {
+      "storage" = "30Gi"
+    }
+    access_modes = ["ReadWriteOnce"]
+    persistent_volume_source {
+      nfs {
+        path   = "/mnt/main/ollama"
+        server = "10.0.10.15"
+      }
+    }
+  }
+}
+
+resource "helm_release" "ollama" {
+  namespace = "ollama"
+  name      = "ollama"
+
+  repository = "https://otwld.github.io/ollama-helm/"
+  chart      = "ollama"
+
+  values = [templatefile("${path.module}/values.yaml", {})]
+}
+
+# Web UI
+resource "kubernetes_deployment" "ollama-ui" {
+  metadata {
+    name      = "ollama-ui"
+    namespace = "ollama"
+    labels = {
+      app = "ollama-ui"
+    }
+  }
+  spec {
+    replicas = 1
+    selector {
+      match_labels = {
+        app = "ollama-ui"
+      }
+    }
+    template {
+      metadata {
+        labels = {
+          app = "ollama-ui"
+        }
+      }
+      spec {
+        container {
+          image = "ghcr.io/open-webui/open-webui:main"
+          name  = "ollama-ui"
+          env {
+            name  = "OLLAMA_BASE_URL"
+            value = "http://ollama:11434"
+          }
+
+          port {
+            container_port = 8080
+          }
+          volume_mount {
+            name       = "data"
+            mount_path = "/app/backend/data"
+          }
+        }
+        volume {
+          name = "data"
+          nfs {
+            path   = "/mnt/main/ollama"
+            server = "10.0.10.15"
+          }
+        }
+      }
+    }
+  }
+}
+
+resource "kubernetes_service" "ollama-ui" {
+  metadata {
+    name      = "ollama-ui"
+    namespace = "ollama"
+    labels = {
+      app = "dashy"
+    }
+  }
+
+  spec {
+    selector = {
+      app = "ollama-ui"
+    }
+    port {
+      name        = "http"
+      port        = 80
+      target_port = 8080
+    }
+  }
+}
+
+
+resource "kubernetes_ingress_v1" "ollama-ui" {
+  metadata {
+    name      = "ollama"
+    namespace = "ollama"
+    annotations = {
+      "kubernetes.io/ingress.class" = "nginx"
+    }
+  }
+
+  spec {
+    tls {
+      hosts       = ["ollama.viktorbarzin.me"]
+      secret_name = var.tls_secret_name
+    }
+    rule {
+      host = "ollama.viktorbarzin.me"
+      http {
+        path {
+          path = "/"
+          backend {
+            service {
+              name = "ollama-ui"
+              port {
+                number = 8080
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/modules/kubernetes/ollama/values.yaml b/modules/kubernetes/ollama/values.yaml
new file mode 100644
index 00000000..1a893909
--- /dev/null
+++ b/modules/kubernetes/ollama/values.yaml
@@ -0,0 +1,18 @@
+ollama:
+  gpu:
+    # -- Enable GPU integration
+    enabled: false
+
+    # -- GPU type: 'nvidia' or 'amd'
+    type: "nvidia"
+
+    # -- Specify the number of GPU to 1
+    number: 1
+
+  # -- List of models to pull at container startup
+  models:
+    - llama3
+
+persistentVolume:
+  enabled: true
+  existingClaim: "ollama-pvc"