I believe we discussed this on IRC, but for the record: there is no "small change" to Apache that is acceptable to the Apache maintainers. https://issues.apache.org/bugzilla/attachment.cgi?id=23051 contains a small change that solves the problem for us, but because it's not a fully general solution (https://issues.apache.org/bugzilla/show_bug.cgi?id=39727#c31 explains why), it has not been applied. Solving the problem in Apache requires allowing each server in a chain of servers to append a _different_ string to the ETag. Or, it requires making transfer-encoding work in a setup with chained HTTP intermediaries, so that we don't have to do all these content-encoding hacks.
I believe we discussed this on IRC, but for the record: there is no "small change" to Apache that is acceptable to the Apache maintainers. https:/ /issues. apache. org/bugzilla/ attachment. cgi?id= 23051 contains a small change that solves the problem for us, but because it's not a fully general solution (https:/ /issues. apache. org/bugzilla/ show_bug. cgi?id= 39727#c31 explains why), it has not been applied. Solving the problem in Apache requires allowing each server in a chain of servers to append a _different_ string to the ETag. Or, it requires making transfer-encoding work in a setup with chained HTTP intermediaries, so that we don't have to do all these content-encoding hacks.