Keeping a failed backend out until manual intervention
If one of our backends fails, we’d like to have it stay out of the rotation until a human manually returns it to service, even if its health checks start to pass again. Possible?
-
Part of the issue here is that an origin's health is from the perspective of each and every POP.
e.g. internet weather events may make a POP briefly unavailable.
This means that sometimes an origin may appear to be unavailable from Aukland but be fine from Seattle, so how to manage this gets a bit more complicated.
-
Hello Peter, Thanks. you are right it can be quite complex. The thing if one of our backend if never used by Fastly (as maybe down from internet) we don't have any way to know it. Same question as well from lhankins : Is there a way to put a backend offline/online easily ? (I guess we need to edit/deploy VCL in all case. no ? Thanks, Alex
-
The thing if one of our backend if never used by Fastly (as maybe down from internet) we don't have any way to know it. Same question as well from lhankins : Is there a way to put a backend offline/online easily ? (I guess we need to edit/deploy VCL in all case. no ? joe root hr@ Online Assignment help
-
hello Peter, I agree but you need to connect to server and have access to it. If one operation do it as server level it's manual operation and can do a mistake. If with your new products Load Balancing https://www.fastly.com/products/load-balancing/ it's possible to do it in the interface ? Thanks, alex
-
There's an API for the dynamic servers. You could remove them from the pool: https://docs.fastly.com/api/dynamicservers#intro
Please sign in to leave a comment.
Comments
7 comments