Use below google sheet for the conversion of heart rate zone setting for both Garmin and Suunto Watches

p.s. This heart rate zones setting is recommended by a trail running group in Hong Kong called - The Peak Hunter. Which is specially tunned for trail running.

The new iOS version update is coming next week. The most interesting new feature for existing homepod mini owners will be the temperature and humiliation metrics. Yes, you've paid for those extra sensor all the time until now it  is becoming useful.

All you need to do is update all the devices connected to the Home to the latest xOS ( tvOS, iOS and homepod mini's firmware) to version 16.3. And then from the environment, you will see those new metrics.

Below is the list of tasks that required for creating the Kubernetes Cluster on the Cloud.

1. Apply for the Oracle Cloud and using the free-tier (life time free) to create 3 Ubuntu Linux Machines using the Ampere CPUs

   • First machine should be the Kubernetes Control plane with 2 CPU, 12G RAM and 50G Storage.
   • 2nd and 3rd machines are the worker nodes with 1 CPU, 6G RAM and 75G Storage.
   • Remember to put all machines in the same virtual network
   • Generate the SSH key and put the public key to each machine. So you can have easy acess to all machines.

2. Install the Docker services in each Machine. You can refer to the link below
   Setup Docker in Ubuntu

3. Install the Kubernetes services in each Machine, with below link
   How to Install Kubernetes Cluster on Ubuntu 22.04
   Few Points to note:

   1. You can skip the installation of the Dockers, as it had been done previously
   2. replace any CPU architecture (i.e. from amd64 to aarch64)
   3. After you installed the first machine, remember to add the firewall rule to allow port 6443. Add rule to iptables

4. Continue the installation steps mention in linuxtechi, which is to join the worker notes with the follow command and setup the network so that it can make those nodes to ready.

   sudo kubeadm join k8smaster.example.net:6443 --token <token>
   --discovery-token-ca-cert-hash <cert>

   **Don't worry about the and as it will provided by the installation script from Kubernetes

Happy New Year!! 先嚟同大家拜個早年。諗咗好耐，有乜方法可以總結2022 呢？其中一個可以share 嘅，就係我係2022 年聽咗咩歌。不過要留意，我嘅2022 歌單唔係話啲歌係2022 出版, 只係真係我個人覺得今年聽過覺得好聽嘅歌. 有興趣就一齊開啲歌嚟聽下啦.

JY Song List 2022

Yeah! I did it, this is my last achievement of 2022 -
Certified Google Cloud Architect.
And hopefully, more can come next year!!

Photo credit to Wan's family!!

Notes for Google Cloud Certificated Cloud Architect

Following are the 7 operation services:

1. Monitoring - Monitoring agent collect CPU, Network and Process metrics. This creates alerts as well. We can also configure monitoring
agents for third party apps.
2. Debugger - Troubleshoot production application code issues.
3. Logging - Logs from Services, has query capability. Use router sink to send logs to GCS, Pub/Sub or BigQuery
5. Profiler - monitor production for memory leak, reduce cost and code performance.
6. Trace - Troubleshoot latency issues
7. Error reporting - Send notification when error occurs from log files.

• Stackdriver /GoogldCloud Logging provides you with the ability to filter, search, and view logs from your cloud and open source application services. Allows you to define metrics based on log contents that are incorporated into dashboards and alerts. Enables you to export logs to BigQuery, Google Cloud Storage, and Pub/Sub.
• Reference

About health Check

For a health check on http (port 80) you don't need to configure nothing in the server.
If the resource you are checking isn't publicly available, you must configure the resource's firewall to permit incoming traffic from the uptime-check servers. See Getting IP addresses to download a list of the IP addresses. If the resource you are checking doesn't have an external IP address, uptime checks are unable to reach it.

NoSQL

NoSQL is well for:
Developers are working with applications that create massive volumes of new, rapidly changing data types ג€" structured, semi-structured, unstructured and polymorphic data.

Cloud Dataflow

Cloud Dataflow is a fully-managed service for transforming and enriching data in stream (real time) and batch (historical) modes with equal reliability and expressiveness -- no more complex workarounds or compromises needed.

BigQuery

• BigQuery is designed for large-scale processing of tabular data, for OLAP.
• Tables can be time-partitioned and set to be expired at n days
• JobUser can only grant in Project level while DataView can run query job.

Cloud SQL:

To reduce no. of queries, set the memcache service level to dedicated. Create a key from the hash of the query, and return database values from memcache before issuing a query to Cloud SQL.

Google Cloud Bigtable

Google Cloud Bigtable is a scalable, fully-managed NoSQL wide-column database that is suitable for both real-time access and analytics
workloads.
Good for:
✑ Low-latency read/write access
✑ High-throughput analytics
✑ Native time series support
Google Cloud Bigtable is a scalable, fully-managed NoSQL wide-column database that is suitable for both real-time access and analytics workloads.

Good for:
✑ Low-latency read/write access
✑ High-throughput analytics
✑ Native time series support
Reference: https://cloud.google.com/storage-options/

Type of Database in google:

• OLTP - Cloud Spanner & Cloud SQL
• OLAP - Big Query
• NoSQL - Filestore and Big Table

DataLab

DataLab is used more for data science scenarios like interactive data analysis and build ML models. However, it does not provide anomaly detection OOTB.

Deployment Manager

Deployment Manager is used to automate the process of provisioning infrastructure

Datastore

Each entity in a Datastore mode database has a key that uniquely identifies it. The key consists of the following components:

• The namespace of the entity, which allows for multitenancy
• The kind of the entity, which categorizes it for the purpose of queries
• An identifier for the individual entity, which can be either
• a key name string
• an integer numeric ID
• An optional ancestor path locating the entity within the database hierarchy

An application can fetch an individual entity from the database using the entity's key, or it can retrieve one or more entities by issuing a query
based on the entities' keys or property values.
References:
https://cloud.google.com/datastore/docs/concepts/entities#datastore-datastore-named-key-python,
https://cloud.google.com/datastore/docs/concepts/entities#datastore-datastore-named-key-python

Dataproc

• Google Cloud Dataproc is a fast, easy-to-use, low-cost and fully managed service that lets you run the Apache Spark and Apache Hadoop
• Cloud Dataproc provisions big or small clusters rapidly, supports many popular job types, and is integrated with other Google Cloud Platform services, such as Google Cloud Storage and Stackdriver Logging, thus helping you reduce TCO.
  Dataproc runs Hadoop, so many kinds of jobs are supported automatically. When you create a cluster with Dataproc, the following technologies are configured by default:

• Hadoop
• Spark
• Hive
• Pig
  Only use preemptible nodes for jobs that are fault-tolerant or that are low enough priority that occasional job failure won't disrupt your business.

Identity-Aware Proxy

=> application level authentication (https://cloud.google.com/iap/docs/concepts-overview)
If you want to start using Google Cloud resources but want to retain their on-premises Active Directory domain controller for identity
management, use Google Cloud Directory Sync to synchronise Active Directory usernames with cloud identities and configure SAML SSO.

Sequence of creating MIG from custom image:

1. Create a custom image from the existing disk.
2. Create an instance template from the custom image.
3. Create an autoscaled managed instance group from the instance template

Cloud VPN

Cloud VPN gateways and tunnels are regional objects, not global

Cloud Filestore

Cloud Filestore is a scalable and highly available shared file service fully managed by Google. Cloud Filestore provides
persistent storage ideal for shared workloads. It is best suited for enterprise applications requiring persistent, durable, shared storage
which is accessed by NFS or requires a POSIX compliant file system.
reference: https://cloud.google.com/terms/services-20180724

Google Kubernetes Engine

To enable the GKE cluster to automatically add/remove nodes based on CPU load => Configure a HorizontalPodAutoscaler with a target CPU usage. Enable the Cluster Autoscaler from the GCP Console.
Deploy each microservice as a Deployment (not POD). Expose the Deployment in the cluster using a Service, and use the Service DNS name to
address it from other microservices within the cluster.

Resilience testing

resilience testing can be done by Schedule a disaster simulation exercise during which you can shut off all VMs in a zone to see how your application behaves

About SSD and storage

• Local SSD cannot be used for boot volume (because its Non-persistent ) and always used for temporary data storage.
• pd-ssd ( Performance (SSD) persistent disks) is persistent and can be used as boot disk.
  Reference: https://cloud.google.com/compute/docs/disks#localssds

• Regional persistent disk is a storage option that provides synchronous replication of data between two zones in a region

http response:

Implement retry logic using a truncated exponential backoff strategy when response code is 5xx and 429.

Using the Cron service provided by App Engine, publish messages to a Cloud Pub/Sub topic. Subscribe to that topic using a messageprocessing utility service running on Compute Engine instances.

Instance Template

• When doing rolling update, can choose between Proactive and Opportunistic ( The MIG applies an opportunistic update only when you manually initiate the update on selected instances or when new instances are created. New instances can be created when you or another service, such as an autoscaler, resizes the MIG)

====
GitHub cheat sheet reference:
 https://github.com/blqthien/Google-Certified-Architect-exam-resources/blob/master/GCP-Products-Cheat-Sheet.md

返到英國就要立即準備倫敦馬。因為咁啱預著公司有project 要上，所以無得請假, 唯有盡量捉緊時間放鬆同拉筋。公司係東倫敦，同expo 都算近，我就揀咗星期四lunch time 去。我都係第一次坐Elizibert line, 又真係好新好乾淨，同其他underground 差太遠。

去到Expo, 會場唔太多人，不過N 記啲貨都係俾人搶購緊。其他大會參展俾子又同栢林EXPO 差唔多喎，例如HOKA 同ON Running, 所以搵完亞Jan… 個名就走啦😜。

到比賽當日，其實仲係有啲唔慣跑六大係會由屋企出發。不過咪以為係屋企出發似好方便. 其實個路程係超轉折, 起步點係喺倫敦東 面嘅greenwich，而我就住係西面，仲要咁啱呢幾日火車公人罷工，班次縮減左. 由西去東，要轉三次車，加埋上上落落車站，我要早2 個鐘頭出門口，先可以趕得切。仲有呀，大會就話比賽當日，所有underground同overground 都係唔洗錢. 不過我轉嚟轉去都係坐火車，跟本就對我無用。

去到起點，就比起栢林馬冷清. 因為佢分咗三條路線起步，而每條路線又再分咗起步組別，變咗每一組起步人數唔多. 又咁講，其實係好事嚟嘅，起馬寄行李去厠所上線都流水作業，非常順利。

兩個馬拉松比較，倫敦馬好似多啲人扮鬼扮馬，我問其中一位點解要咁著，佢就話自己今年無乜點練到，所以就搵啲公仔野嚟扮下，咁就可以有另種投入。開始跑後沿路都仲有好多其他唔到角色，有犀牛，有Dory，有Big Ben, 仲有full gear 嘅真消員. 另一個比較係跑嘅路比起栢林馬係窄啲，仲要可能因為倫馬無時間limit, 所以好多人跑跑下會係路中心停落嚟或者慢慢行，仲有啲人係一字排開咁行，又或者話停就停，要成日都打醒精神唔可以跟車太帖，唔係就好易撞到前面.

我同之前一樣都set 咗Heart rate (HR) zone alert, 希望可以做返同栢林馬差唔多pacing 完成就好滿足。去到 10公里之後就發覺唔得啦，油缸無油就係無油，對腳真係無力, 原本係諗住開始時唔好太快，等去到後啲先慢慢加返快. 點知真係諗下好啦. 10K後係真係想快都無得快，HR 同pacing 原全都上唔到. 不過叫咗老婆同屋企人接放學，又唔可以遲得太過份. 所以唯有放棄唔睇錶，放鬆去跑 easy 過easy 嘅easy run. 目的係跑完而唔洗行。

沿路打氣不絕，等別係過tower bridge 同最後5 公里入central london，所有人出曬嚟打氣咁，超熱鬧. 終點設計都好有心思, 經過BIG BEN 要先跑向白金漢宮再轉去另一條大馬路the mall衝線... 咁衝線相嘅背景就會係白金漢宮，因為個背景實在太有意思，所以就俾錢買相啦.

移民後 又搵工，又中covid 所以停操咗一年 . 為咗唔可以俾自己停落嚟，所以舊年年底開始有咩比賽都會報咗先。一來真係唔覺得有咁好彩，二來，倫敦同栢林兩個都係6 大馬拉松之一，所以我同好多人一樣，會習慣報曬兩個先。點知.... 真係咁好彩地中曬😜，但係真係無為意係兩個賽係隔得咁近, 今年一月知道栢林馬中咗，再到今年三月收到倫馬話我中埋, 再check 日子, OMG, 係連馬. 不過同自己講，栢林馬可以練下做下時間. 倫馬係自己屋企嘅比賽, fun run 就得。咁樣，就開始咗呢個星期嘅奇妙之旅.

栢林馬我估好多人都分享咗，所以想由另一個方向分享下。就係我係點48 小時快閃栢林馬。

首先，由屋企出發, 最平嘅方法由倫敦去栢林一定係坐budget airline. 我就㨂咗easy jet, 為咗盡量唔請假同可以用最短時間來回栢林同倫敦，我就㨂咗星期六朝早8 點幾 由GATWICK airport 上機🛫。計過如果嗰朝call 車去, 要至少£80, 仲未計番嚟嗰轉，咁就決定揸車去泊long stay parking, £48 搞掂，泊咗車就用接駁巴士往返機場. 個半鐘機就到栢林機場🛬.

栢林馬最好係比賽期間，市內免費坐公共交通公具, 所以交通費又慳返。Download 定張電子車飛，就可以就咁行入車廂好方便。

去到EXPO同班長會合，行咗一陣然後就去airbnb 攞房，再去同班長、Joanne 佢地食飯。食完飯就返去訓覺。

我係最後一組出發，又揀咗唔洗寄行李，所以第二朝唔洗太早起身。係麥記食咗早餐，再攞定個包上車就去起步集合點。栢林坐地鐵搵路 Google Map 都好可靠。但其實係月台都預到好多一睇就知會去馬拉松嘅跑友，所以要去邊個月台上車，幾時轉車，主要都係跟其他人睇電話都可以慳返。

跟住指示去到起步點集合嗰時，KIPCHUGI 都已經跑到一半，仲話再快過對上一次個WR。大家都起興奮. 比起其他比賽，栢林馬嘅起步都算做有心思，好似UTMB 起步咁，每一組時起步前，大會主持都會同大家一齊拍手，而拍手嘅方法同用嘅音樂，每組都唔同好好玩。

到我嗰組起步，沿路可以睇到好多栢林嘅特色建築 ，好多都好高同好宏偉。跑嘅通常都係三四條行車線咁闊嘅馬路，係人同人之間穿插時都唔太難. 為咗唔令自己第一隻跑爆咗而跑唔到倫敦嗰場。我都係跟住先前係隻錶度set 咗嘅HR Zone 跑, keep 住個zone 跑，就唔會爆. 不過真係操練唔夠，20K 後想加快少少，隻錶己經開始出alert. 咁無計啦，唯有放鬆easy 跑。最後完成到，仲要唔係想像嘅差時間，所以已經好滿足。

跑完第一件事係即刻攞啤酒飲，呢隻無酒精嘅啤酒又幾好飲喎。之後就返去airbnb 沖涼大覺訓，食麥記再再訓多震。我回程班機係朝早7 點機，照計應該要5 點鐘到機場. 但免費車飛零晨3 點就會失效. 所以為咗慳返夜晚嗰轉的士錢，我就決定早啲退房，夜晚11 點就出機場。去到機場其實已經收曬工，係個角落頭搵張長櫈訓覺，訓到機場開返就入閘上機返屋企。咁就結束咗我個 快閃栢林馬之旅。

This is the photo of my lovely teammates

PS5 終於有得upgrade ssd 啦. 真係由三月過嚟之前買定呢條SN850 2TB NVMe. 等足半年, sony 終於可以俾你裝自己嘅SSD 入部機度啦.

仲有, 如果而家上去Amazon 睇下, 同一model 足足俾我之前睇個價貴咗成 £100. 諗返轉頭, 真係無乜幾河咁有眼光

#ssd讀speed要每秒讀速5500MB以上

仲有要加埋heatsink嘅. heatsink唔可以太厚十零£己經ok