ip-tables workaround

.github/workflows/k8s-snap-integration.yaml

@@ -39,6 +39,13 @@ jobs:
           sudo lxd init --auto
           sudo usermod --append --groups lxd $USER
           sg lxd -c 'lxc version'
+      # Docker sets iptables rules that interfere with LXD.
+      # https://documentation.ubuntu.com/lxd/en/latest/howto/network_bridge_firewalld/#prevent-connectivity-issues-with-lxd-and-docker
+      - name: Apply Docker iptables workaround
+        shell: bash
+        run: |
+          sudo iptables -I DOCKER-USER -i lxdbr0 -j ACCEPT
+          sudo iptables -I DOCKER-USER -o lxdbr0 -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
       - name: Build k8s-dqlite
         run: |
           make static

.github/workflows/performance.yaml

@@ -50,12 +50,13 @@ jobs:
           sudo lxd init --auto
           sudo usermod --append --groups lxd $USER
           sg lxd -c 'lxc version'
-      - name: Ensure lxd network traffic flows by removing docker if installed
+      # Docker sets iptables rules that interfere with LXD.
+      # https://documentation.ubuntu.com/lxd/en/latest/howto/network_bridge_firewalld/#prevent-connectivity-issues-with-lxd-and-docker
+      - name: Apply Docker iptables workaround
+        shell: bash
         run: |
-          if command -v docker >/dev/null 2>&1; then
-            echo "Docker is installed, purging it"
-            sudo apt-get purge -y docker-engine docker docker.io docker-ce docker-ce-cli containerd runc
-          fi
+          sudo iptables -I DOCKER-USER -i lxdbr0 -j ACCEPT
+          sudo iptables -I DOCKER-USER -o lxdbr0 -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
       - name: Download latest k8s-snap
         run: |
           sudo snap download k8s --channel=latest/edge --basename k8s

pkg/kine/server/mett.txt

@@ -0,0 +1,79 @@
+
+
+- initial brainstorm workshop
+
+OPEA 30 minutes intro, micro services spin up agents show architecture.
+
+graph, databases
+
+afterwards canonical part of the workshop   
+
+having the deployment on microk8s, 
+neo4j, vanilla example but extend it to the edge
+
+infrastructure provided no problem
+
+
+90 minutes
+
+30/40 minutes 
+
+
+environment, instructions, EKS, 
+
+
+cloud nighttag intel developer cloud
+ipcs there
+aipc
+
+access to the environment
+
+idc account 
+
+
+hardware: aipc 
+
+nvidia gpu 
+gaudi 
+
+run RAG with opensearch 
+
+cpu only with gpu its fast without its slow like 1minute per response
+
+
+run RAG architecture 
+
+
+access gpu
+
+pvcs or gaudis checking what arch we have
+
+run RAG on gaudi accelerator
+
+on cloud they dont have nvidia
+
+what generation zion
+
+24.04 ubuntu 16 cores 1 soket 
+
+aipc, gpu 
+training on dater center on the edge
+
+probably nothing on gaudi more on zion
+
+
+share overall architecture
+
+chatqnA
+
+opensearch, 
+
+graph rag deployed in our environment
+
+
+OPEA - OpenPlatform for Enterprise AI (linux foundation project)
+--> RAG - Retrieval Augmented Generation
+
+ChatQnA - Chat Question and Answer bot 
+
+