This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(259.20000000000005, 68%, 35%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ET%3C/text%3E%3C/svg%3E)
Team,
In eastus, various promotional offer machines (suffix _promo) are available which offers less cost in comparison to their standard version. What are the advantages and disadvantages of them using in production. I am writing one disadvantages here but other than that is there anything else.
My main concern is this link which says if aks found machine unhealthy after reboot then it will delete/reimage/recreate it. But once the promotional offer ends how will it recreate.
I would be grateful if anyone can help here
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(179.20000000000002, 30%, 27%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESK%3C/text%3E%3C/svg%3E)
Hi @Tanul
I would suggest not to use promo virtual machine for now in production.
Refer the below link for your better understanding.
For more in depth understanding please raise support ticket on this.
Cheers!!
@Shivam Kumar Singh , Thank you for sharing those details. But can you help me with the answer on that recreation thing because some of other teams applications are already running in production with promo machines.
What will happens if a situation of automatic node recreation comes in front of AKS. I would be grateful if you can help me here. I need to involve other people for raising a support ticket which will be time consuming..
Hi @Tanul , you need to check the General Availability of these Promo Virtual machine to tackle the situation of automatic node re-creation failure if the promo virtual machine is not available after the October 31st, 2020 for NC series promo, NV series promo and the H series promo VM.
I didn't find any GA date announced for these promo virtual machines.
I hope this will helpful for you.
@Shivam Kumar Singh , The standard versions of these machines are already available. But, consider a scenario if promo machines are not available anymore then AKS will keep on making a request for the same machine which might create an issue.
I can monitor it manually and increase the size of the node by attaching the standard version of the same. But, how AKS will replicate all the data from that old set to new one because for aks it is still a different machine. There has to be some identifiers which convince aks to consider the new machine and deploy the old artifacts/data like any deployment etc. which could have been already deployed in old machine.
In such cases, while replacing a node manually what are the things we have to consider so that aks can consider and easily reimage the new machine it with the existing data.
@Tanul , I would suggest you will add standard version of your desired VM and apply Taints and Tolerations concept to control the deployment of pods in any specific worker node.
@Shivam Kumar Singh , Sorry Shivam I didn't get the concept here. How taint and toleration can help in this case.
Consider a scenario that tomorrow, microsoft plans to decommission the standard version of any machine and by mistake we are using the same machine. In such case also, aks will start throwing errors that node is unavailable.
Does aks automatically replicates the resources sprayed with tolerant of that failed node on this new node(just attached with same taint).
As far as I know K8's automatically schedule the pods evenly on different nodes but with this approach it will become necessary to tolerant all the pods we are using and ask every team who is going to use this cluster to add tolerant on there deployments.
@Shivam Kumar Singh , Typing here as the word limit was over above
Are you sure this is the only possible way through which aks understand node and replicate resource in case of disaster recovery. Because, it would definitely increase the overhead for a developer to remember the tolerant and more of a problem for me to calculate the available ram and cpu of the nodes so that accordingly, I can confirm/distribute the tolerant value to the user before every deployment.
Aren't we challenging the scheduling feature of K8's here.
In addition, at least we can use the promotional machines in UAT environment for saving the money. UAT's are always replica of prod and they also have same resources. How to handle the situation there. Admin can add taint on nodes but how one can edit yaml of all the deployments as each of that have different type of ram and cpu usage. How to rebuild and evenly distribute them using tolerant.
@Shivam Kumar Singh , Hello Shivam, good evening..
Did you get chance to work on this issue. Please let me know if you find any update here..
Thank you
Have a nice day
Hi @Tanul , Did you tried to add a second agentpool to the cluster and choose the new VM type you want for that new agentpool and deploy your application. Then you can also delete the old promo vm pool if eveything is working fine.
For more details, see https://learn.microsoft.com/en-us/azure/aks/use-multiple-node-pools
I hope this will resolve your issue.
Cheers!!
@Shivam Kumar Singh , oh ok.. So basically creating another node pool let say with same labels and aks will leave the other one and consider the new one..
In short, i think whole devops is working on the basis of selectors and left over aks handles and redploy things with the help of ETCD server..
Thanks a lot shivam for the help. Happy weekend. Tk cr :)
Welcome @Tanul :)
Please mark this as an answer if you got your solution :)