Setup with Ansible

Quick Start

INFO

https://github.com/spantaleev/matrix-docker-ansible-deploy/blob/master/docs/quick-start.md

Prerequisites

INFO

https://github.com/spantaleev/matrix-docker-ansible-deploy/blob/master/docs/prerequisites.md

1. Install deps

# On Control Node
pipx install --include-deps ansible # 
pip install passlib # 
pipx install rust-just # or sudo apt install just
sudo apt install pwgen # pwgen -s 64 1

2. Configure firewall
3. Configure DNS settings

ess-helm

1. install pg on dedicated server: install pg

2. set up database for synapseset up database for mas

3. enablee password

   # ~/17/data/pg_hba.conf
   host all all 10.33.12.111/32 scram-sha-256

   # in postgres user
   /usr/pgsql-17/bin/pg_ctl reload

4. ess-helm get started

5. Config elementWeb/synapse ...

6. Install

helm upgrade --install --namespace "ess" ess oci://ghcr.io/element-hq/ess-helm/matrix-stack -f ~/ess-config-values/hostnames.yaml -f ~/ess-config-values/tls.yaml -f ~/ess-config-values/postgresql.yaml  --wait

# rocky linux
dnf install git tar # required for install helm

WARNING

kubectl create namespace ess: 报错 The connection to the server localhost:8080 was refused

cp /etc/rancher/k3s/k3s.yaml .kube/config
kubectl create namespace ess

如果 helm upgrade 失败，请检查日志：kubectl get events -n ess --sort-by=.metadata.creationTimestamp

可能是自部署的 postgres 无法联通

pg

# /home/postgres/17/data/postgresql.conf
listen_addresses = 'xx.xx.xx.xx'

[
   "postgres",
   "-c",
   "max_connections=500",
   "-c",
   "shared_buffers=2611MB",
   "-c",
   "effective_cache_size=7834MB",
   "-c",
   "maintenance_work_mem=652MB",
   "-c",
   "checkpoint_completion_target=0.9",
   "-c",
   "wal_buffers=16MB",
   "-c",
   "default_statistics_target=100",
   "-c",
   "random_page_cost=1.1",
   "-c",
   "effective_io_concurrency=200",
   "-c",
   "work_mem=1337kB",
   "-c",
   "huge_pages=off",
   "-c",
   "min_wal_size=1024MB",
   "-c",
   "max_wal_size=4096MB",
   "-c",
   "max_worker_processes=16",
   "-c",
   "max_parallel_workers=16",
   "-c",
   "max_parallel_workers_per_gather=4",
   "-c",
   "max_parallel_maintenance_workers=4"
]

服务器调整

# /etc/sysctl.d/xxx.conf

# 文件描述符限制
fs.file-max=1000000

# this sets the max, so that we can bump the JVB UDP single port buffer size.
net.core.rmem_max=10485760
net.core.netdev_max_backlog=100000

sudo sysctl -p