SRE Simplified - Introduction
What are the SRE concepts, like Service Level Agreement and Reliability?
The challenge is to explain it in simplified manner, without deep discussions on different monitoring strategies.
In this first video I tried to just to scratch the surface.
Wireguard VPN Overview
WireGuard is a relatively new protocol. It claims to be greatly simplified and faster comparing to alternatives, it uses UDP as primary transport and encapsulates encrypted traffic from IP layer. Let's investigate if it's true!
Some differences help to stand out of other VPN solutions:
- Routing as a part of Peer configuration
- QR code for simplified configuration using camera on mobile devices. iPhone, iPad and other things, you'll love it!
- Site-to-site or client-to-server connections are not very different from the configuration perspective
- Supports DNS as part of configuration for Peer
- Easy to implement routing for all traffic or for only for selected networks
I was able to connect Mikrotik, FrtizBox 6690, macos, iPad and iPhone using WireGuard with little efforts.
How to suppress alerts in Prometheus for non-working hours
When it comes to the topic how Prometheus and Alertmanager work together and how to restrict alert notification to fire only on working hours, typicall solution is to apply workaround to avaluated alert rule expession like this
and ON() (hour() < 19 and hour() > 8) and ON() (day_of_week() > 0 and day_of_week() < 6)
The full example of such rule is the following
- alert: DummyWorkhoursAlert
expr: 1 and ON() (hour() < 19 and hour() > 8) and ON() (day_of_week() > 0 and day_of_week() < 6)
for: 5m
annotations:
identifier: "{{ $labels.exporter }}/{{ $labels.queue }}"
summary: "Buyerportal: Dummy alert for working hours {{ $labels.queue }}"
description: "* don't do anything, dummy alert *"
But the proper solution is to apply mute_time_intervals, as per example below
Blogging using Emacs with org-mode and Hugo
I use org-mode extensively with org-roam for keeping my knowledge and journals together and working as my brain dump destination. This blog serves kinda the same purpose, so why not to combine these things under Emacs umbrella?
Blogging from Emacs boils down to two options.
TLDR; Hugo can render org file natively, but support of features is quite limited, or use org as source that rendered to markdown. The latter brings issues of content syncing from origin to end result.
Install Ruby 3.1.4 on macOS
In case ruby can’t find OpenSSL 3 headers, use this way.
brew install ruby-install
ruby-install ruby 3.1.4 -- --with-openssl-dir=$(brew --prefix openssl)
CloudFlare Tunnel Terraform
How to provision CloudFlare tunnel using Terraform
CloudFlare Tunnel can be useful to use reliable alternative to ngrok when you need to expose your application running locally to the outside world.
The following example exposes my application locally running on port 3000 to the Internet on the hostname https://app.prokopov.me
Prerequisites
- Terraform CloudFlare module v4
- CloudFlare API key with Account:Tunnel permissions.
- Locally installed CloudFlare CLI utility cloudflared https://developers.cloudflare.com/cloudflare-one/connections/connect-apps/install-and-setup/installation/
- Application running locally on port 3000
How does it work?
- cloudflared CLI is an agent running locally and connected to CloudFlare cloud.
- DNS record of type CNAME is created, pointing to the CloudFlare cloud.
- CloudFlare does the routing magic!
Terraform part
resource "cloudflare_tunnel" "main" {
account_id = "777414c2d4e87234087ebac4685e7df6"
name = "tunnel-to-app"
secret = random_id.main.b64_std
}
resource "cloudflare_tunnel_config" "main" {
account_id = "777414c2d4e87234087ebac4685e7df6"
tunnel_id = cloudflare_tunnel.main.id
config {
warp_routing {
enabled = true
}
ingress_rule {
hostname = "app.prokopov.me"
service = "http://localhost:3000"
}
ingress_rule {
service = "http_status:404"
}
}
}
resource "cloudflare_record" "main" {
value = "${cloudflare_tunnel.main.id}.cfargotunnel.com"
proxied = true
name = "app"
type = "CNAME"
zone_id = cloudflare_zone.main.id
}
Local tunnel part
- Find generated token for resource cloudflare_tunnel.main
TOKEN=$(terraform show -json | jq -r '.values.root_module.resources[] | select(.address=="cloudflare_tunnel.main").values.tunnel_token')
- Use token
cloudflared tunnel run --token=${TOKEN} tunnel-to-app
Mikrotik Terraform
It is in general very good idea to manage infra configuration as a code. Unfortunately, Mirkotik terrafrom support is basic, as OSS driven.
Nevertheless, I appreciate author for effors.
Here is an example how to use it with Hashicorp Vault.
Commands
export VAULT_ADDR=http://vault_address:8200
export VAULT_TOKEN=<token>
terraform init
terraform plan
Example
main.tf
provider "vault" {}
data "vault_generic_secret" "main" {
path = "common/mikrotik/nexus-home"
}
provider "mikrotik" {
host = data.vault_generic_secret.main.data["address"] # Or set MIKROTIK_HOST environment variable
username = data.vault_generic_secret.main.data["username"] # Or set MIKROTIK_USER environment variable
password = data.vault_generic_secret.main.data["password"] # Or set MIKROTIK_PASSWORD environment variable
tls = false # Or set MIKROTIK_TLS environment variable
ca_certificate = "/path/to/ca/certificate.pem" # Or set MIKROTIK_CA_CERTIFICATE environment variable
insecure = true # Or set MIKROTIK_INSECURE environment variable
}
// /ip address
// :put [find where address="192.168.88.1/24"]
// *1
// terraform import mikrotik_ip_address.lan '*1'
resource "mikrotik_ip_address" "lan" {
address = "192.168.88.1/24"
comment = "LAN Network"
interface = "ether2"
}
// uncomment on release https://github.com/ddelnano/terraform-provider-mikrotik
# resource "mikrotik_firewall_filter_rule" "https" {
# action = "accept"
# chain = "forward"
# comment = "Web access to local HTTP server"
# connection_state = ["new"]
# dst_port = "443"
# in_interface = "ether1"
# in_interface_list = "local_lan"
# out_interface_list = "ether3"
# protocol = "tcp"
# }
terraform.tf
Group Greeting
There are lots of options to greet collegue or close ones with cards.
This one was recommended by one of the coworkers.
The things impressed me recently
Impressive things and points of interest
Self-Improvement
https://theholisticpsychologist.com/ - mental self-healing of issues from the Past
Technologies assessment
https://www.nomadproject.io/ - K8s is not only one option for container orchestration. Looks good!
https://www.boundaryproject.io/ - Remote access management from Hashicorp
https://www.ambient.run/ - WebAssembly Game engine targeting browsers
https://www.playstation.com/de-de/ps-vr2/ - New Sony VR2 headset.
Technology adoption
https://docs.docker.com/build/buildkit/ Docker buildkit and it’s advanced caching techniques.
Why is IT support so hard
As the IT support business we want to keep our users happy, they need to use the software with no interruptions. Though the share of the incidents in the tickets is still 30% no matter what.
So why is IT support is so hard these days? Why do we still have the incidents despite all the progress IT industry did so far?
Long story short this is because of the software complexity, that causes incidents and security issues.