sueboy

[轉]OVS (Open vSwitch) 心得楊金龍

https://www.facebook.com/dominic16y/posts/1378435818935569

今天跑去台中聽了OVS (Open vSwitch)講座，收獲頗豐(真的值回車票價了!)，這個東西我自己的研究可能要花好幾個月的時間，還不一定能抓對方向或抓到重點，講師今天一個下午就把所有關鍵的問題都提到了，比如說你linux bridge 用的好好的幹嘛要轉到ovs來，這是大家都會關心的問題。

底下一些重點整理:
1.linux bridge效能很好，沒有什麼特殊的需求的話，不用轉到OVS去。
2.linux bridge效能的極限，kernel 跑不滿10G的頻寬(大約7G/s左右)，若你不用10G的網路的話，linux bridge很好用。
3.OVS原生的效能比linux bridge效能差
4.OVS必須要另外加DPDK才能夠跑滿10G頻寬。(大約可跑到15~20G/s)
5.OVS單台機器沒有什麼用，它的功能跟linux bridge一樣
6.OVS必須要多台，並加上controller才能發揮它真正的功能
7.controller 推薦ONOS、OVN
8.OVS支援所有L2的功能(和靜態路由L3功能)，它要使用controller之後，就可以不用STP，可充份利用所有頻寬，也不怕looping所造成的廣播風暴。

srs rtmp forward (push) out || pull in push out

master 192.168.105.20 srs.conf

isten 1935;
max_connections 1000;
srs_log_tank console; #file; #console;
srs_log_file ./objs/srs.log;
daemon off; #on or delete this line
http_api {
enabled on;

listen 1985;

}

http_server {

enabled on;

listen 8080;

dir ./objs/nginx/html;

}

stats {

network 0;

disk sda sdb xvda xvdb;

}

vhost __defaultVhost__ {

ingest livestream {

enabled on;

input {

type stream;

url rtmp://xxx.ooo.xxx.ooo/live/nna1

}

ffmpeg ./objs/ffmpeg; #if no build, just install then link -s

engine {

enabled off;

output rtmp://127.0.0.1:[port]/live?vhost=[vhost]/nna2 #this push back this server

}

forward 192.168.105.21:1935;

gop_cache off;

queue_length 10;

min_latency on;

mr {

enabled off;

}

mw_latency 100;

tcp_nodelay on;

}

slave 192.168.105.21 srs.conf

listen 1935;

max_connections 1000;

srs_log_tank file; #console;

srs_log_file ./objs/srs.log;

#daemon off;

#http_api {

# enabled on;

# listen 1985;

#http_server {

# enabled on;

# listen 8080;

# dir ./objs/nginx/html;

#stats {

# network 0;

# disk sda sdb xvda xvdb;

vhost __defaultVhost__ {

}

==========finsih===========for test======

obs push video to master rtmp://192.168.105.20/live/obs1

master player rtmp://192.168.105.20/live/obs1

slave player rtpm://192.168.105.21/live/obs1

***srs pull rtmp nna1 then push localhost nna2 ***

check rtmp://xxx.ooo.xxx.ooo/live/nna1 first

master player rtmp://192.168.105.20/live/nna2

slave player rtmp://192.168.105.21/live/nna2

PS:

1、debug just

a. daemon off

b. srs_log_tank console;

then see screen have any error msg. Some info like restart ffmpeg.

2、if slave can't play, just use port :1935

============
go-oryx

linux usb boot windows

https://www.pendrivelinux.com/yumi-multiboot-usb-creator/

https://www.pendrivelinux.com/universal-usb-installer-easy-as-1-2-3/

[轉]在Ubuntu 使用 Lets Encrypt 與 Nginx

https://blog.technologyofkevin.com/?p=591

see http://sueboy.blogspot.tw/2017/11/nginx-proxy-pass-best-practices.html
http://sueboy.blogspot.tw/2017/11/nginx-proxy-pass-best-practices-2-for.html

docker db Volumes

http://dockone.io/article/128

http://container42.com/2014/11/18/data-only-container-madness/

http://dockone.io/article/129

在volume产生时，是docker run的准备阶段（create），而执行entrypoint.sh则是在启动阶段（start）

clustercontrol install finish

ssh-copy-id better use root

ssh-copy-id finish

Reboot clustercontrol server

======sysbench======

[sysbench]

select

sysbench --num-threads=128 --report-interval=3 --max-requests=0 --max-time=300 --test=/usr/share/sysbench/tests/include/oltp_legacy/select.lua --mysql-table-engine=innodb --oltp-table-size=5000000 --mysql-user=root --mysql-password="E<8r)^t-E<8r)^t-" --oltp-tables-count=1 --db-driver=mysql --mysql-host=192.168.xxx.xxx --mysql-port=3306 prepare

sysbench --num-threads=128 --report-interval=3 --max-requests=0 --max-time=300 --test=/usr/share/sysbench/tests/include/oltp_legacy/select.lua --mysql-table-engine=innodb --oltp-table-size=5000000 --mysql-user=root --mysql-password="E<8r)^t-E<8r)^t-" --oltp-tables-count=1 --db-driver=mysql --mysql-host=192.168.xxx.xxx --mysql-port=3306 run

Oltp read/write

* 产生数据
sysbench --num-threads=128 --report-interval=3 --max-requests=0 --max-time=300 --test=/usr/share/sysbench/tests/include/oltp_legacy/oltp.lua --mysql-table-engine=innodb --oltp-table-size=5000000 --mysql-user=root --mysql-password="E<8r)^t-E<8r)^t-" --oltp-tables-count=1 --db-driver=mysql --mysql-host=192.168.xxx.xxx --mysql-port=3306 prepare

* 执行
sysbench --num-threads=128 --report-interval=3 --max-requests=0 --max-time=300 --test=/usr/share/sysbench/tests/include/oltp_legacy/oltp.lua --mysql-table-engine=innodb --oltp-table-size=5000000 --mysql-user=root --mysql-password="E<8r)^t-E<8r)^t-" --oltp-tables-count=1 --db-driver=mysql --mysql-host=192.168.xxx.xxx --mysql-port=3306 run

*清除
最後面改成 cleanup

Operating System User clustercontrol

https://severalnines.com/docs/requirements.html#operating-system-user

firebase hosts cloudflare

ERR_TOO_MANY_REDIRECTS

Or

Error 526

SSL change Full or Full (strict)

just try

go gae telegram bot telegrambot

https://blog.sean.taipei/2017/05/telegram-bot
https://github.com/cortinico/telebotgae

Use this token to access the HTTP API:
445067055:AAG8Lz1EL---------------------------------

．https://api.telegram.org/bot<Token>/<Method Name>
https://api.telegram.org/bot445067055:445067055:AAG8Lz1EL---------------------------------/getMe
https://api.telegram.org/bot445067055:445067055:AAG8Lz1EL---------------------------------/getUpdates

．https://api.telegram.org/bot<Token>/sendMessage?chat_id=<Chat_id>&text=<Message>

chat_id 使用者送訊息給telegram bot後，去getUpdates，就會有chat id

https://api.telegram.org/bot445067055:AAG8Lz1EL---------------------------------/sendMessage?chat_id=265----------&text=Hello+World

．https://api.telegram.org/bot<Token>/setWebhook?url=<Url>

刪除是用 deleteWebhook
取得Webhook資料 getWebhookInfo

https://api.telegram.org/bot[API_KEY]/setWebhook?url=https://[PROJECT-ID].appspot.com

https://api.telegram.org/bot445067055:AAG8Lz1EL---------------------------------/setWebhook?url=https://xxx---bot-17------.appspot.com

inscapsula cdn ddos free plans

https://www.incapsula.com/pricing-and-plans.html

最近才知道 Incapsula 蠻有名的，有免費試用喔

免費的額度是

頻寬 5Mb/s

1個月2T以內

[轉]async 、 await es6 promise javqscript

https://hackmd.io/s/Bko9gUGYZ

ip firewall filiter raw input forward

https://www.mobile01.com/topicdetail.php?f=110&t=3205444&p=574#5738

ip firewall filter是在nat動作後

ip firewall raw在nat動作前

當封包透過nat進來後會分兩個路徑: 1.提供路由器(Router)自用 2.給區域網路的電腦使用

若要指定給路由器(Router),進入chain使用input
若要指定給區域網路的電腦,chain則一律用forward

一般病毒防護都著重在區網的桌機,所以要在firewall filter進行過濾,chain設為forward即可.

RouterOS路由器的smb伺服器預設是關閉的,所以若沒開啟不太需要再新增一筆chain=input封鎖項目.

/ip firewall raw操作:
在nat之前的意思即,封包在進入路由器前就先判定有敵意.
所以在這操作即不管這封包最終是路由器使用,還是給區網的電腦操作一律都先行丟棄,這樣明白吧XD

input/forward都是在nat後的動作,所以在firewall raw看不到這兩個chain.

prerouting是在nat前的動作,所以在firewall filter看不到這個chain.

[轉][RouterOS] 上傳流量爆頻寬 DNS

https://tlcheng.wordpress.com/tag/rb750gr3/

go gae mux static html css js images

1、裝官方gcloud工具
2、裝python 2.7
3、到官方工具目錄 ..........\Local\Google\Cloud SDK\google-cloud-sdk
3-1、設定path //執行 install.bat 會有說明

source

在原本開發go的目錄執行下面指令及可，但要特別注意出現套件未安裝的話 go get github.com/xxxx 裝一裝就行了

dev_appserver.py app.yaml //這裡要特注意，因為是local測試，會出錯！因為init()問題，如果是用flexible就不會，但又要改app.yaml

gcloud app deploy

gcloud app browse

======
gcloud init //reset project id

注意：gae flexible environment vs standard environment
https://cloud.google.com/appengine/docs/the-appengine-environments
所以要用 standard environment

====================
https://www.slideshare.net/takuyaueda967/gaegoweb

elk Elasticsearch Logstash and Kibana fortigate ubuntu

https://www.rosehosting.com/blog/install-and-configure-the-elk-stack-on-ubuntu-16-04/

https://www.elastic.co/guide/en/logstash/current/configuration.html

https://dotblogs.com.tw/supershowwei/2016/05/25/185741

install finish

1、/etc/logstash/conf.d/ put some logstash conf

2、ubuntu have logstash listen error, so nano /etc/logstash/startup.options
LS_USER = root

3、/usr/share/logstash/bin# ./system-install reuse LS_USER for config

注意：

mutate {
add_field => {
"logTime" => "%{+YYYY-MM-dd} %{time}"
}

[轉]Docker工具新手入門　用指令熟悉容器生命週期

http://netadmin.com.tw/article_content.aspx?sn=1708010003&jump=2

aws linux partition resize

http://docs.aws.amazon.com/zh_cn/AWSEC2/latest/UserGuide/ebs-expand-volume.html#recognize-expanded-volume-linux

1、AWS console EC2 resizse

2、login EC2

2.1、 lsblk get info to disk size. Is resize ok?

2.2 resize2fs /dev/xvda1

If 2.2 finish, then resize faild. follow 2.3

2.3 parted /dev/xvda

2.3.1 parted> print all -> get disk real size
2.3.2 parted> resizepart
2.3.3 parted> 1
2.3.4 parted End?>put size is 2.3.1
2.3.5 parted>exit

then 2.2 again. If ok then 2.1 check age.

PS：
2.3.4

End?> -1

-1 最大磁碟空間