Josef Bacik posted on Mon, 21 Apr 2014 07:55:46 -0700 as excerpted: [Near the bottom, point #4 immediately before conclusion.] > You still have to post-process merge to make sure, but you are far more > likely to merge everything in real-time since you are only changing the > sequence number every once and a while instead of for every new sequence > number. Much of this is above my head, which is OK as I'm not a btrfs dev, but... ...only changing the sequence number... for every new sequence number? Something's wrong with that. No WONDER the system live-locked! =:^) My above-my-head guess is that in place of for every new sequence number there, you meant for every delayed ref. Meanwhile, thanks for putting this all down. I already said I don't understand much of it, but I certainly have a better appreciation for the complexities of snapshot-aware-defrag. I hadn't considered what qgroups could do to it at all! -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
