You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently our CI is just using a single zone to get spot requests. It seems that the spot capacity is pretty low for m5.xlarge so our instances get killed pretty quickly.
There are some solutions to this problem we could consider.
Spot Block - this is the most obvious solution so we block the spot request for a given time frame. In our case one hour would be enough
Spot Fleet - with spot fleet we will use a broader set of availability zones and instance types that fullfill our requirements to avoid using AZ and instance type with low capacity
Currently our CI is just using a single zone to get spot requests. It seems that the spot capacity is pretty low for m5.xlarge so our instances get killed pretty quickly.
There are some solutions to this problem we could consider.
for now we deactivate spot instances with 9b45cc8
D2iQ Jira Ref: D2IQ-66089
The text was updated successfully, but these errors were encountered: