Closed captncraig closed 6 years ago
:cry: though I'm not surprised.
The general latency for operations I ran into in the handful of tests I ran while writing the original PR was 1 minute at most (generally 30-45 seconds).
I have no issue with simply removing the wait code and documenting the oddity with an eye toward potentially adding an alternate waiter if it ends up becoming confusing and/or a problem for folks.
Any volunteers? R53 is a very popular provider and we don't want to diverge too far from their SDK.
I can take it if no one else gets to it first, but I won't have time until early next week. If no one else has officially claimed it by then, I'll start in on it.
It's really just removing some code. I'll take care of it real quick.
Status?
@tlimoncelli it seems it has been done in commit ccf28349ceac9a32fce5d406ce777b6371d285a3.
Great! I'm closing this.
On updating the aws package with
govendor fetch github.com/aws/aws-sdk-go/...
in the vendor directory, dnscontrol fails to build:It appears that the sdk removed the waiter package. It breaks our code that waits for nameserver changes to propagate. Alternate functionality may be available with the this package (but the warnings there don't help me, and I can't seem to find the right way to use that), or by calling GetOperationDetail yourself in a loop.
I'm really not sure what the best practice is for this. The api seems real odd to me. I'm also not opposed to just removing the wait code if its too much hassle.
I don't have any domains with them, or I'd be very interested in testing actual latency to see how big a concern this waiting actually is. If the worst case is we make a few redundant corrections because they lag, I may be ok with that.
cc @maclennann