1. If you are using the balanced orientation which is a mix of cost orientation along with the availability orientation, it means it will always launch instances that have lowest pricing at the moment in the Az's along with the longest duration without any disruptions. Spot instances service provider usually make choices based on the lowest pricing and long lasting instances. But this does not mean these service provider will evenly spread between AZ's i.e. balanced orientation is not always balanced distribution.
2. You can usually change this by selecting the availability orientation but this options narrows down the possibility of long continuity instance types in case volatility increases so choose that option with consideration.
3. Now there can be issue in case your subnet is not having sufficient capacity to create more ip address. This can happen when there are not enough free addresses available in the subnet to satisfy the requested number of instance.
4. Also if the instances has already reached the maximum number of instances in elasticgroup configuration than also new instances won't be spawn so in this case you should monitor the value i.e. current instances is always 25% lower than maximum instances in case you receive some spike and more containers are required to be deployed due to which more servers are required to spin up.
2. You can usually change this by selecting the availability orientation but this options narrows down the possibility of long continuity instance types in case volatility increases so choose that option with consideration.
3. Now there can be issue in case your subnet is not having sufficient capacity to create more ip address. This can happen when there are not enough free addresses available in the subnet to satisfy the requested number of instance.
4. Also if the instances has already reached the maximum number of instances in elasticgroup configuration than also new instances won't be spawn so in this case you should monitor the value i.e. current instances is always 25% lower than maximum instances in case you receive some spike and more containers are required to be deployed due to which more servers are required to spin up.
0 comments:
Post a Comment