How to Simplify the Developer Experience with AKS

Key Takeaways

hello everyone my name is prakaran from the Azure customer experience team and today I'm joined by Tom and tats from the Azure AKs engineering team to talk more about some of the cool new capabilities added to the extension since the last episode and with many interactions with our customers we were able to capture those valuable feedbacks and translate them into new features within the AKs vs code extensions such as reconcile cluster comparing two clusters aborting the last operation and capturing granular TCP dump and eras a tool and all new create new cluster uh I now invite Tom and tads to talk about the new features added to the vs code extension uh Tom uh do you want to introduce yourself and talk about these features sure yeah I'm Tom I'm engineer on the AKs team um yeah so our General approach with uh these features is trying to find ways to make um common tasks as easy as possible to remove friction and to really allow developers to continue with their workflow from within the vs code environment they're already familiar with um create is a quite straightforward allows you to create a cluster very quickly using some standard setup um compare um again does what it sounds like it allows you to compare two different clusters so you can see you can figure one slightly differently is there something you might want to make a change to um reconcile abort um hopefully you're familiar to um AKs users but the idea is basically um if you have a cluster that is not in a succeeded State you can reconcile it so it basically takes the desired State and applies that to the running cluster um abort again is to stop the most recently started operation so if you realize you've made a mistake you don't want to do what you just submitted stop that TCB dump is a phenomenal tool that people use for troubleshooting network issues it can be somewhat complex to set up and to run yourself um so our idea here is to really provide that interface um right in within BS code and finally the Eraser tool um again does what it sounds like it allows you to erase um older images um from your nodes and keep your system secure and clean as possible with that I will turn it over to tats to actually show us some demos nice hi proo hi Tom thank you so much this is absolutely right and thanks Tom for putting it such a nice way so I'll start in the same order what Tom said um compare AKs cluster this is one of the user uh us feature via our repo and I think the idea is that a lot of time people have two clusters one is working one is not working or sometimes user want to compare to Cluster configuration so for that sometimes user compare use Json object copy it from the portal put it in the file and then compare the two file using third third party tool we are not saying don't use the third party tool but here user by just simple click can use this tool to compare two cluster in this case I've just compared cluster one and cluster two which I created for this demo and here I can see what the provisioning state differences it can come handy when the scenario is that why this cluster is in succeeded space whereas another one is in updating and then user can easily see why with the configuration difference that isn't handy tool and under the hood we do all the heavy lifting for the specific feature moving forward another use when there are um blue green deployments right identical clusters and customers wants to see the configuration drift between these clusters yes yes indeed indeed like depending on the yeah depending on the experience of the user I mean as you correctly pointed out braa uh people can use it to their expert lb the second which Tom correctly pointed out was the new create cluster ux which is pretty simple here we have upgraded user can create in a one click without leaving their favorite editor uh and sometimes Engineers are in that mode that they are doing some coding and they want to create a new Dev test cluster now user can by simply selecting their Resource Group uh here I selected My Demo group and let's let's name it um MF and then let's select Central us and then boom it goes it is hand shaking with the arm call Via R SDK and now it will send a request to create the cluster create cluster does take time so since you guys have seen the uh ux I'll move on to the next which is reconcile and abort operations this is again one of the customer feature which came to us via I think prha you bya our open source repository so in this case let's say for example I have a msft cluster to which is in a good State and I decided to do a manage cluster operation say rotate cluster certificate and instantly saying yes to it instantly I went to do some another work and realized oh no I don't want to do that operation in my cluster so I go to my show properties page back again and hit abort this is abort operation will only show when there is an operation going on which involv the starting of the cluster and boom see the request which was getting sent by vs code via our arm call which is an SDK is cancelled because of the abort click and you can see that operation is now canceled and this is a special with longer running um activities such as update or upgrade um if you have a pdb on your cluster and you realize you know what this upgrade is not going to finish in a timely manner I want to make some changes first um you can cancel the upgrade make your changes do what you need to do and then rerun the uh that long running operation so you're not just waiting indefinitely for something that potentially will take quite a while awesome and and this is true for any cluster management operation right yes yes and uh one of the things I came to know is that reconcile was quite was used quite a few times when your cluster was in a bad state right so having this kind of functionality at the user do step also get Advanced user as a new user to actually explore if they want to reconcile try reconcile rather than having the whole cycle and waiting for a few days to get back I think they can use this tool to do that so reconcile recommendation is definitely to use reconcile to take a cluster that's in a uh non uh successful State versus say stopping the cluster and starting the cluster um which could create other problems the reconcile is a much safer and uh less impactful change to make and now you can see after hitting recall it is doing a board uh updating the cluster so we'll leave that at this state and move on to other one because at some point of time the reconcile will happen and this cluster will again come on the Happy State we can check this out at the end of this demo let's move to um and we can check this new create cluster as well later on at the end of the demo let's move on to the uh TCP dump one of the feature we created again from the OSS repo uh feature request for capturing TCP dump traffic for Network diagnosability currently there is a document which explains how to collect TCP dump which is four to five steps for the user to uh gen to create to capture their cap files download it and then they can send if they understand how to use it they can debug otherwise they send it to the CSS Engineers or the expert support Engineers here we have taken the functionality to the doorstep into one click so essentially if user has a cluster and bunch of nodes they will select right click on their cluster and they will select something called collect TCP dumps in this case I have my debug pod already installed otherwise user will have something called create button appearing here in those scenarios they all they need to do is click on the create and soon after the little bit of a progress they will see this uxx we have added filters but for now I want to run this without filters so that we can show the amount of traffic I have done without filters for the audience these are the filters we have added so let's let's go ahead and start with the default capture so I'll I'll click on the start button so imagine all that five to six steps user need to do is now ux via vs code extension and all they need to do is click on the create pod and then sit back and click on the start and count until how far they want to you know get the traffic for the last 30 minutes they can wait for 30 30 seconds and um let's say I want to capture for like 10 seconds and I will stop now what happens at the background is now all the heavy lifting is done by this vs code EXT ention to at this point stop the TCP dump capturing and then present it at this point we can actually download it from the node to your two users local machine here I'm going to select downloads so just a just a thought tats here you know if we can add an upload button where users can upload it to a storage account and then share it with the support team that really Bridges the Gap um of sharing the capses with the support team and the filter looks very cool uh I think that this will help uh customers to capture very granular level of uh the traffic avoiding all those noises and you know uh just capture uh what is required for that scenario indeed and uh thank you for the advice I would highly recommend to go to the open source repo and definitely which at the end of the this presentation I can show it to you so as you can see after clicking download it has downloaded to to the location I said and now I'm all I'm going to do is capture this since I know how to read the uh cap file I'll go in here and I'll do TCP dump d r paste the file name and you can see that easy it is instead of going through the steps creating your own deployment deploying it it's in about 30 seconds we have TCP dump in your local machine you can now use other was kind of tools to provide for the filter or user can like you mentioned braa and Tom previously this these kind of early filters can help in filtering out the people who know or the expert label depending on the label they are or this could this cap file can be shared with the CSS and support Engineers so now that initial experience of capturing that TCP dump is super simple and just in a couple of clicks user can get it now moving on to the last feature which is create eraser tool which is kind of again used by some of the uh customers for actually cleaning up the garbage collection for their uh nodes which so eraser tool is wellknown tool which is used for cleaning up the unused images so all user need to do is right click select their cluster right click and select run eraser image cleanup and all it does is checks the user's properties if your cluster is in good State once this is done mightbe I'll check I'll take the other one as well razor there you go in this case uh it'll reassure recheck that user really wanted to now the both appear I'll just take the one um yes I want to clean up unused images and off it goes it deploys the Eraser cluster Again One Click feature all the heavy lifting done for the user and um at this point use engineer can focus in their developer experience in their tool and so on so forth and this all experience is just one click for them and it has successfully run against this cluster so when customer wants to clear up some disk space you know due to uh the way they use the image P policy it can you know fill up very quick and sometimes it puts the you know operating system into a readon mode so this will help avoid all those um hiccups in the cluster it's very cool and just to close the loop you remember at the start create cluster we created ttom pra at open at msft it's now created so I can right click go to the show property page and I can see it's in succeeded state so the operation we were waiting for long running operation it has created as well so yeah create cluster is already created and that's about it at the end I would really encourage users to uh kindly go to our repo we have a discussion board uh user can add their suggestions here or they can go to the repo itself um which is vs code AK tools and open the feature request via issues and we keep a very close eye on the repo um just to fill the last Gap if user don't know how to uh download this extension they can go to their extenstion manager and look for AKs and upon that they will see Azure Community Services that's where since I've already downloaded it appears as downloaded it's a very light extension that gets installed instantly right and then we're going to share those repo the links to the repo in the description to the video right tats yes absolutely we will it's awesome and the previous videos as well where we have explained the other features so these are the latest few few features and that's uh that's about it thank you Dad thanks for this cool demo and thanks uh Tom for joining us today and I guess you enjoyed watching this uh demo and the cool new features that are available through vs code extensions I encourage you to go explore download the extension share your feedback and your comments thank you so much for watching this