gistfile1.txt

13:44 paravoid: fyi, I'm going to recommend to not use container sync
13:45 paravoid: apart from the bugs I've found it's unbelievably slow, partly because it syncs per container and because it's horribly inefficient at that: it opens a new http connection for every PUT for example
13:45 paravoid: and takes 30 seconds on each run to read all of the container databases
13:46 paravoid: also, I tried running it from two container servers and they sync the same files -- quite the inverse of what you'd expect of parallelization
13:46 paravoid: all in all, it seems like an experimental feature useful for some edge cases of remote-initiated uploads that is not suited for our full blown datacenter replication needs