Just a blog - Karim Elatov
I wanted to try out a simple use case of having an agent automatically create draft emails for my unread messages in my inbox. After checking out a couple of options I decided to try out kagent . My initial idea was to take this approach kagent uses the gmailmcp server to get the unread emails kagent then uses ollama (I setup ollama in the past ) to create the content for the draft message kagent…
I decided to install OPNSense on my firewall which has been running pfSense for a couple of years. The majority of pfSense configurations map directly to OPNsense, ensuring high functional parity between the two platforms (they are both using freeBSD under the hood). Here are the settings I applied: Disable beeps: System > Settings -> Miscellaneous -> Disable the startup/shutdown beep Configure…
I wanted to convert the rest of my VMs to use EUFI on proxmox. Here are the steps I followed. Create a dedicated /boot/efi partition I had two machines and they had different partition schemas. Shrinking / with LVM The first machine looked like this: > sudo fdisk -l /dev/sda Disk /dev/sda: 50 GiB, 53687091200 bytes, 104857600 sectors Disk model: QEMU HARDDISK Units: sectors of 1 * 512 = 512 bytes…
I ran into Ingress NGINX Retirement: What You Need to Know and I decided it’s time to migrate to the Gateway API . I decided to use the istio gateway controller, since I was planning on using the auth methods. Also this encouraged me: Gateway API Benchmarks Installing Gateway CRDs Since I want to use tlsroute , I need to install the experimental CRDs: > kubectl apply --server-side -f…
Back in the day I tried deploying open-webui as a container on my old proxmox server and the performance was abysmal. So during the holiday season I decided to get a new server with a GPU (I picked HP Z2 Mini G1a ) and try running open-webui one more time. Migrate VMs to new Proxmox Server I was planning on decommissioning the original server so I didn’t want to enable HA and live migrate the VMs.…
I recently used release v2.27.0 of kubespray to upgrade my kubernetes cluster and I ran into an interesting issue. Control Plane Upgrade Failure I just kicked off the upgrade as I usually do: > ansible-playbook -i inventory/home/hosts.yaml -b upgrade-cluster.yml And it failed on the control plane node, during the drain the node hung: TASK [upgrade/pre-upgrade : Set if node needs cordoning]…
iwlwifi module issues I came back home one day and my machine running gentoo was having issues connecting to the internet. I looked at the logs and I saw the following: Jan 05 17:44:16 nuc systemd-networkd[645]: wlp0s20f3: Gained carrier Jan 05 17:44:16 nuc wpa_supplicant[878]: wlp0s20f3: CTRL-EVENT-CONNECTED - Connection to xx:xx:xx:xx:xx:xx completed [id=0 id_str=] Jan 05 17:44:16 nuc…
After switching to rook-ceph I needed to update my velero configuration to make sure I can take backups of my PVCs. In the process I ended up learning a lot about the velero backup flow. Velero Snapshot Data Movement By default velero doesn’t backup data to a remote location (this is briefly discussed in this issue ). It just creates a snapshot and if a backupStorageLocation is configured it will…
With the recent news about openEBS getting archived by CNCF OpenEBS: Lessons We Learned from Open Source , I decided this was the time to try out another storage provider and I decided to try rook-ceph . I’ve heard so much about ceph in the past but I never really tried it… so this time around I decided to give it a chance. I have a pretty small k8s cluster with just 3 VMs running on a Proxmox…
I did my regular kubespray upgrade and after the upgrade some things stopped working. Here are some issues I ran into. DNS failures After the upgrade I started seeing the nodelocaldns pods show a bunch of permission denied messages: As a quick test I decided to disable that feature : enable_nodelocaldns: false And then perform an upgrade, but after that nodelocaldns was still running. Looking at…