To avoid redundancy and inconsistencies, Reverse DNS lookup should be centrally adressed in the markdown at one place and then not be mentioned further. I think it is actually not that surprising that we don't want this to happen, so can we just generally say that for all the address and domain attributes, instrumentations SHOULD NOT do a DNS lookup on their own (neither forward nor reverse) but capture the result of the application/libraries lookup or not set the corresponding attribute at all (or in some cases set it to whichever of the set {IP, name} is available). This is not only due to performance but also to avoid reporting inaccurate information for multi-IP hosts by default.
To avoid redundancy and inconsistencies, Reverse DNS lookup should be centrally adressed in the markdown at one place and then not be mentioned further. I think it is actually not that surprising that we don't want this to happen, so can we just generally say that for all the address and domain attributes, instrumentations SHOULD NOT do a DNS lookup on their own (neither forward nor reverse) but capture the result of the application/libraries lookup or not set the corresponding attribute at all (or in some cases set it to whichever of the set {IP, name} is available). This is not only due to performance but also to avoid reporting inaccurate information for multi-IP hosts by default.
(Related #206)