>>>>> "NeilBrown" == NeilBrown <neilb@xxxxxxxx> writes: NeilBrown> On Thu, 16 Jul 2015 11:07:34 -0400 "John Stoffel" <john@xxxxxxxxxxx> NeilBrown> wrote: >> >> So.... how will wrapping of sequence numbers be handled in the cache >> metadata on disk? They can't grow for ever... NeilBrown> Can't they? NeilBrown> A 64bit number incremented every millisecond can grow for half a NeilBrown> billion years. As far as I'm concerned, that is forever. LOL... that is true. But I'm trying to just think about edge cases and point out some possible pitfalls in case of corruption or errors. NeilBrown> But even if we used 32bit numbers, it is very easy to NeilBrown> handle looping providing you have few than half of the NeilBrown> numbers in use at any time. So as long as we don't have NeilBrown> more than 2 billion active transactions in the log at once, NeilBrown> there is no problem. The kernel does some internal time NeilBrown> keeping using "jiffies" which, on 32bit systems, cycle NeilBrown> about every fortnight. A small amount of care is enough NeilBrown> that this doesn't cause problem. All I wanted to make sure, was that this was thought of and handled. I like the idea of randomly picking a STARTING number, so that we hopefully exercise all the corner cases properly early on in the process. John -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
