All gists from anduin

anduin / Docker install tzdata

0 likes

0 forks

1 files

Last active 1719250124

1	RUN DEBIAN_FRONTEND=noninteractive apt-get update && \
2	apt-get install -y tzdata && \
3	echo "Etc/UTC" > /etc/timezone && \
4	ln -fs /usr/share/zoneinfo/UTC /etc/localtime && \
5	dpkg-reconfigure -f noninteractive tzdata

anduin / Azure Service Bus to Kusto

0 likes

0 forks

2 files

Last active 1725699790

1	using System.ComponentModel;
2	using System.Data;
3	using System.Text;
4	using Aiursoft.Canon;
5	using Azure.Identity;
6	using Azure.Messaging.ServiceBus;
7	using Kusto.Data;
8	using Kusto.Ingest;
9	using Microsoft.Extensions.DependencyInjection;
10	using Microsoft.Extensions.Logging;

anduin / CUDA monte carlo estimate pi

0 likes

0 forks

1 files

Last active 1717217849

monte carlo estimate pi

1	#include <iostream>
2	#include <curand.h>
3	#include <curand_kernel.h>
4
5	__global__ void calculate_pi(int *count, unsigned long seed) {
6	int idx = threadIdx.x + blockIdx.x * blockDim.x;
7	curandState state;
8	curand_init(seed, idx, 0, &state);
9
10	float x = curand_uniform(&state);

anduin / fake service.yml

0 likes

0 forks

1 files

Last active 1715092025

Make any docker service fake

1	# Fake entrypoint for the service. Always sleep.
2	entrypoint: ["sleep", "infinity"]
3	# Fake healthcheck for the service. Always return true.
4	healthcheck:
5	test: ["CMD", "true"]

anduin / Low Saturation Fix

0 likes

0 forks

3 files

Last active 1742359572

这部分内容是通用的用于诊断 Substrate 数据中心机器部署失败的方法。可以从宏观上找到核心问题。

准备工作区：立刻打开两个DMS，两个OSP和一个Kusto Explorer。
识别：识别有故障的范围，是版本还是Ring。在OSP检查此Ring趋势图。检查Substrate版本历史，确认其版本类型（Dogfood、Daily）。

我正在检查 ________ 版本，范围是 Deploy Ring _________，其部署状况是 ______，版本类型是 ______。

anduin / Ceph deploy notes

0 likes

0 forks

1 files

Last active 1712420160

Ceph deploy notes

这里将在 3 台 node 上安装 ceph。

三台 node 的 hostname 分别是：

node1
node2
node3

三个 node 全部都插一块新的大硬盘。假如是 /dev/sdb 。

anduin / Ubuntu ISCSI

0 likes

0 forks

1 files

Last active 1710779939

Ubuntu mount ISCSI server

1	# Ubuntu install ISCSI Server
2	# Assuming the IP of the server is 10.0.0.1
3
4	# Mount a disk as /dev/sdb first
5	apt install tgt -y
6
7	# Edit /etc/tgt/conf.d/iscsi.conf
8	echo "
9	<target iqn.2024-03.iscsi.server:lun1>
10	backing-store /dev/sdb

anduin / Mount gluster fs

0 likes

0 forks

1 files

Last active 1710702440

1	# Assuming using /dev/sdb is your new empty disk
2	# Assuming 10.0.0.* is your private network
3
4	# Install and configure GlusterFS. (Run on all nodes)
5	apt-get install glusterfs-server -y
6	systemctl start glusterd
7	systemctl enable glusterd
8
9	# Format the disk and mount it (Run on all nodes)
10	mkfs.xfs /dev/sdb

anduin / 从 Vsphere 切换到 Docker Swarm 的收益

0 likes

0 forks

1 files

Last active 1712559086

省去虚拟机的烦恼
攻击面显著减小。不需要维护那么多虚拟机的密码。
不再需要配置复杂的 Linux 权限，反正在容器里就是 Root。
不再需要配置复杂的路由器和 Linux 的路由表、networkmanager，反正容器开了都是建好的。
不再需要重复运行大量无意义的 Linux 内核，巨大的节省资源
过去，运行三个独立的业务，需要三个 VM。包括缓存，可能至少需要占用 20GB 的主机内存。
迁移到容器后，唯一的内存开销就是容器的进程，每个业务只需要 200MB 内存。

不再需要担心内存泄漏而对服务器定时重启，而是只重启容器即可

anduin / Koel deployment

0 likes

0 forks

1 files

Last active 1710087208

Koel deployment

1	echo "Downloading koel..."
2	sudo apt install tar wget jq
3	latestUrl=$(curl https://api.github.com/repos/koel/koel/releases/latest \| jq -r '(.assets[] \| select(.browser_download_url \| contains(".zip"))).browser_download_url')
4	rm /tmp/koel.zip
5	rm /tmp/koel_ext -rf
6	wget -O /tmp/koel.zip $latestUrl
7
8	echo "Unziping koel..."
9	unzip /tmp/koel.zip -d /tmp/koel_ext