0:38 - High availability and scalability 4:17 - Disaster recovery 6:27 - Kubernetes vs. AWS 👉🏼 For any questions/issues/feedback, please leave me a comment and I will get back to you as soon as possible. Please don't forget to subscribe and give me a thumbs up if you liked my video 😊
Hi Nana - Thank you for wonderful series of presentation and explanation, I have few questions here, 1. How many worker node can control by each master node (what is the recommended number). 2. In cluster, is same worker node can managed by more than one master node? 3. In etcd, how the cluster states are maintained, is there any dedicated process available or etcd itself communicated with each worker node? 4. Each POD should have dedicated service or same service can mapped more than one POD from same server?
Thanks Nana, i request you to please make video on implementing practically how disaster recovery is done in Kubernetes Cluster and how we can recover back the cluster again using the ETCD snapshot.
You are the best Nana !! 3 things: Would you mind putting something up like " how to build HA Kubernetes cluster via Rancher" , "pod deployment and services thru Helm" , and Elastic search deployment in Kubernetes cluster. as always I really love the rich content of your knowledge and presentation. Thanks Again for your time and being gracious !!
Thanks a lot Mathew, really happy to hear such positive feedback about my videos 🙂Great suggestions! I already plan to do videos for Helm Demo and ELK stack in K8s. Will consider Rancher & HA as well for later!
Hello Nana - I have been following your videos for some time now, thank you for creating wonderful content and sharing with us. Could you possibly create a video about DR setup and moving workloads to the DR site in case of disaster in primary K8s cluster? Thanks.
@@sagarmaske8761 They are not the same. But they basically do the same thing, which is manage pod replication. ReplicaSet is actually a replacement for Replication Controller and is used by deployment to manage replicas.
@@sagarmaske8761 Good point, that's why the Replication Controller will become deprecated in the future. So you can think of the ReplicaSet as an improved replication controller.
You never mentioned Docker Swarm. I don't particularly like it but it's a pretty major solution for some people and I don't think I've ever used a service that can scale easier than it.
Yes, this video is specifically for k8s. But I have a comparison video between k8s & Docker swarm if you're interested: th-cam.com/video/9_s3h_GVzZc/w-d-xo.html .What is your experience in advantages and disadvantages of both?
Hi, thank you for the helpful videos. I am moderately new to k8s. It would be really great to see videos on SNAT port exhaustion and how to detect it. I have read and gone through few videos i could find but cudnt get more out of it. The other thing we r perplexed on our applications get deployed on Azure K8s and they hold up for abt 30 mins. Then all serivces degrade in network thru put and cant connect to .net core or services. I am beginning to think we r exhausting socket/ports when pods scale more and more. i am a .net core developer and so much fascinated how k8s is architected. Its one of those infrastructure if we dont do it right it bites back. thanks in advance
Incredible! I have read so much documentation and stuff and nobody could explain how this actually works.... Thank you for the great examples. Keep up the good work! One question though. If you have like 10 nodes workers and 10 websites or apps. If you deploy them all they will be in each node right? Can it be set so each app can use its own node with 1 replica ?
Great video, thanks. I have a question for you: What does happen if I reboot all my etcd nodes simultaneously ? Will I lose all its data ? Are they ephemerous pods ? Second question: Can I use Persistent Volumes to mount etcd data directory to persist its data out of the cluster ?
Great Video ,can not be any simpler and clear explanation with precise diagrams ! well as far as HA is concern Two nodes within Primary DC serves the website content via INGRESS ,however I wanted to have Secondary DC running similar workload and this could be available either in HOT/HOT or HOT/WARM . Just wondering how this will be work in case of Disaster scenario and what needs to be taken care. 1st thing I will ensure data is replicated between two DCs using some storage level mechanism like SAN Replication etc. 2nd I will add GTM ..Global Traffic Manager between two DCs (Global Server Load Balancer from BIGIP F5 or any other brand) . Do I need to create another INGRESS in DR Site and add these two INGRESS Entries under GTM ? In normal scenario User will access GTM and get redirected to one of the INGRESS based on how GTM is configured. If GTM is configured in Round Robin it will be HOT/HOT Setup and if GTM is configured to redirect traffic only to Primary Site then it would be HOT/WARM. My Question is how POD replicas are created in DR Site as part of this setup ? will it be via one DEPLOYMENT or need to create Secondary site setup using another independent DEPLOYMENT ?
Hi Nana, thanks for this excellent series! Question though, you mentioned that ingress balances the request load, is load balanced, and that there's an instance of ingress on each node. Where's the entry point for a request to "my-app"? Is there some sort of Ingress Master that listens for requests to "my-app" communicates with Ingress slaves on each node to determine how to balance the request load? Thanks in advanced for any insights.
Hi Nana, It would be great if you can share your thoughts Assume I have two clusters Cluster 1 at Site 1 (main site) Cluster 2 at Site 2 (DR site) Active-Active DR A request comes at Cluster 1 to the API. The API tries to connect to the database. Assume all replicas of DB in cluster 1 are unreachable for some reason. Can I achieve a failover to the DB in DR site? Generally speaking I want the following Prefer a local component to process a request. If not available failover to the DR site and continue further processing locally at DR and trace the response path back in main site into a component that failed over to a component in DR.
Please can u help me with a doubt Do we need docker volume for the applications if we Deploy it with kubernetes. If we use which volume should we use docker volume or kubernetes volume Please help me with this
Hi! Can you please help me? When I delete a pod and its replacement pod goes ready, it triggers session termination among other live pods.Do you have any idea? What is it about pod uptime that is affecting other pods?
If i have a file server pod, and there are 10 files server agents, where will the file come from from the downloaders pov. The loads balancer or the fike server?
K8s doesn't help in managing the underlying server infrastructure. So in this case, you'll have to restart the node yourself and re-join it to the cluster.
0:38 - High availability and scalability
4:17 - Disaster recovery
6:27 - Kubernetes vs. AWS
👉🏼 For any questions/issues/feedback, please leave me a comment and I will get back to you as soon as possible.
Please don't forget to subscribe and give me a thumbs up if you liked my video 😊
Still not understand the difference between K8 and AWS autoscaling
I just started my Devops career 3 wees ago. Thank god I found your tutorial. You are explaining it very well.
what wees ?
Hi Nana - Thank you for wonderful series of presentation and explanation, I have few questions here,
1. How many worker node can control by each master node (what is the recommended number).
2. In cluster, is same worker node can managed by more than one master node?
3. In etcd, how the cluster states are maintained, is there any dedicated process available or etcd itself communicated with each worker node?
4. Each POD should have dedicated service or same service can mapped more than one POD from same server?
Thanks Nana, i request you to please make video on implementing practically how disaster recovery is done in Kubernetes Cluster and how we can recover back the cluster again using the ETCD snapshot.
You are the best Nana !! 3 things: Would you mind putting something up like " how to build HA Kubernetes cluster via Rancher" , "pod deployment and services thru Helm" , and Elastic search deployment in Kubernetes cluster. as always I really love the rich content of your knowledge and presentation. Thanks Again for your time and being gracious !!
Thanks a lot Mathew, really happy to hear such positive feedback about my videos 🙂Great suggestions! I already plan to do videos for Helm Demo and ELK stack in K8s. Will consider Rancher & HA as well for later!
Thanks for adding this simple way to understand kubernetes. Kindly add more videos Elastic DB, kibana, Granfana
Awesome tutorial so far I have ever seen. Thanks for your hard work and for sharing the knowledge. Kudos !!
Thanks Nana, Keep going, your tutorials are awesome. rich content and attractive explanation.
Thanks a lot Mahmoud, glad you enjoyed it :)
Done thanks
Todo take notes in onenote
Again a great video. Thank you.
Thank you :)
Hello Nana - I have been following your videos for some time now, thank you for creating wonderful content and sharing with us. Could you possibly create a video about DR setup and moving workloads to the DR site in case of disaster in primary K8s cluster? Thanks.
Thankyou @TechWorld with Nana . Your teaching skills re really exceptional. Keep on the good work.
Thanks a lot, glad you like it! 🙂
I have a question. Are RepicaSet and replica controller the same?
@@sagarmaske8761 They are not the same. But they basically do the same thing, which is manage pod replication. ReplicaSet is actually a replacement for Replication Controller and is used by deployment to manage replicas.
@@TechWorldwithNana if both of them do the same thing, why need two of them ?
@@sagarmaske8761 Good point, that's why the Replication Controller will become deprecated in the future. So you can think of the ReplicaSet as an improved replication controller.
Such a good explanation nd the presentation, really appreciate u effort. Keep it up great work
Thanks Nina! Really helpful!
You're welcome Drake. Best of luck and stay safe. 🙂
another banger, thank you!
Thanks again Marc!
this is a very very helpful series , thank you
I'm glad this has helped you Mohamed. Please don't forget to share with your friends. 🙂
@@TechWorldwithNana yeah this helped me alot , thank you ♥♥
Good clarity . Awesome!!!
Very nicely explained
You never mentioned Docker Swarm. I don't particularly like it but it's a pretty major solution for some people and I don't think I've ever used a service that can scale easier than it.
Yes, this video is specifically for k8s. But I have a comparison video between k8s & Docker swarm if you're interested: th-cam.com/video/9_s3h_GVzZc/w-d-xo.html .What is your experience in advantages and disadvantages of both?
big thanx very appreciated tutorial.
Hi, thank you for the helpful videos. I am moderately new to k8s. It would be really great to see videos on SNAT port exhaustion and how to detect it. I have read and gone through few videos i could find but cudnt get more out of it. The other thing we r perplexed on our applications get deployed on Azure K8s and they hold up for abt 30
mins. Then all serivces degrade in network thru put and cant connect to .net core or services. I am beginning to think we r exhausting socket/ports when pods scale more and more. i am a .net core developer and so much fascinated how k8s is architected. Its one of those infrastructure if we dont do it right it bites back. thanks in advance
Thank you Nana.
Thank you very much for the video !!
Incredible! I have read so much documentation and stuff and nobody could explain how this actually works.... Thank you for the great examples. Keep up the good work!
One question though. If you have like 10 nodes workers and 10 websites or apps. If you deploy them all they will be in each node right? Can it be set so each app can use its own node with 1 replica ?
Great video, thanks. I have a question for you: What does happen if I reboot all my etcd nodes simultaneously ? Will I lose all its data ? Are they ephemerous pods ? Second question: Can I use Persistent Volumes to mount etcd data directory to persist its data out of the cluster ?
Mam please make a video on multi tenancy kubernetes architecture.
Thanks for these great videos Nana! Can you compare kubernetes with AWS serverless setup aka lambda functions?
Great Video ,can not be any simpler and clear explanation with precise diagrams ! well as far as HA is concern Two nodes within Primary DC serves the website content via INGRESS ,however I wanted to have Secondary DC running similar workload and this could be available either in HOT/HOT or HOT/WARM . Just wondering how this will be work in case of Disaster scenario and what needs to be taken care. 1st thing I will ensure data is replicated between two DCs using some storage level mechanism like SAN Replication etc. 2nd I will add GTM ..Global Traffic Manager between two DCs (Global Server Load Balancer from BIGIP F5 or any other brand) . Do I need to create another INGRESS in DR Site and add these two INGRESS Entries under GTM ? In normal scenario User will access GTM and get redirected to one of the INGRESS based on how GTM is configured. If GTM is configured in Round Robin it will be HOT/HOT Setup and if GTM is configured to redirect traffic only to Primary Site then it would be HOT/WARM.
My Question is how POD replicas are created in DR Site as part of this setup ? will it be via one DEPLOYMENT or need to create Secondary site setup using another independent DEPLOYMENT ?
u are the best !! thank u so much
You’re the best 🤙
@Nana Thank you for the valuable content
Very nice videos and explanation. Just curious What do you use to make these slides ?
Hi Nana, thanks for this excellent series! Question though, you mentioned that ingress balances the request load, is load balanced, and that there's an instance of ingress on each node. Where's the entry point for a request to "my-app"? Is there some sort of Ingress Master that listens for requests to "my-app" communicates with Ingress slaves on each node to determine how to balance the request load? Thanks in advanced for any insights.
in baremetal world, how would you declare who is master and who is slave? simply by installing the respective processes?
Which component ensures that if some pod or whole node dies than this information should be updated in etcd? As mentioned on 4:11 in this video.
Awesome!!!
Is it possible to add dynamicaly or manual nodes in the kubernettes cluster,master and worker nodes?
thank you
thank you！
Are you dev ops specialist ? how to practice 8 at home . My company do not have k8 implemented yet but i wanna learn cloud , docker and k8
Hi Nana,
It would be great if you can share your thoughts
Assume I have two clusters
Cluster 1 at Site 1 (main site)
Cluster 2 at Site 2 (DR site)
Active-Active DR
A request comes at Cluster 1 to the API. The API tries to connect to the database.
Assume all replicas of DB in cluster 1 are unreachable for some reason.
Can I achieve a failover to the DB in DR site?
Generally speaking I want the following
Prefer a local component to process a request. If not available failover to the DR site and continue further processing locally at DR and trace the response path back in main site into a component that failed over to a component in DR.
Please can u help me with a doubt
Do we need docker volume for the applications if we Deploy it with kubernetes.
If we use which volume should we use docker volume or kubernetes volume
Please help me with this
Hi! Can you please help me? When I delete a pod and its replacement pod goes ready, it triggers session termination among other live pods.Do you have any idea? What is it about pod uptime that is affecting other pods?
thx nana :)
If i have a file server pod, and there are 10 files server agents, where will the file come from from the downloaders pov. The loads balancer or the fike server?
How to take backup of kubernetes cluster and restore
This is a complex topic with many variables. Can't explain in a couple of sentences, so maybe I'm gonna make a video about it sometime later.
@@TechWorldwithNana
Thanks it would really helpful if you make video soon
HELP!!!! SESSIONS ARE TERMINATED WHEN ANOTHER SAME APPLICATION STARTS FROM OTHER PODS :(
what happen if master is down? would all the nodes goes also down?
If all masters are down, worker nodes will still run, but you won't be able to access them, because you won't have the cluster control plane.
What is the solution if 1 worker node dies. How kubernetes can restore the dead worker node?
K8s doesn't help in managing the underlying server infrastructure. So in this case, you'll have to restart the node yourself and re-join it to the cluster.
@@TechWorldwithNana Thanks for your quick reply.
I got a empty web page . but in view code I can see all the html & js code ?!
Nana please marry me