Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Fri, 2014-06-27 at 10:52 -0400, Allen Chen wrote: > I found an interesting feature from DRBD when I test DRBD compered > with > MD raid1. > I am seeking a "relax" disk backup solution. Of cause, both DRBD and > raid1 can do it. > But I need this feature: if the second node(DRBD) or disk(raid1) is > offline(no disk errors, no nothing,just disconnected), and comes back > later, I need the two nodes(DRBD) or disk(raid1) to be synced as soon > as > possible. > In this case, DRBD takes a few seconds to sync whatever changed, and > raid1 takes hours to sync the whole disk(also needs annually add the > out > of sync disk). > So I want to use DRBD to replace raid1 in this case. > Does it have any issue to run DRBD on the same machine? > > Allen There was a post by Bram Klein Gunnewiek to the list on June 19 with brief instructions on how to setup two DRBD nodes on the same system that synchronize with each other. There have also been several posts today saying that, if you enable write-intent-bitmap for a Linux MD RAID array, it will only resync the changed parts of the drive. This is the first sentence from the link that Roland (rf at q-leap.de) replied with, from the wiki page describing the feature: >When an array has a write-intent bitmap, a spindle (a device, often a >hard drive) can be removed and re-added, then only blocks changes since >the removal (as recorded in the bitmap) will be resynced. Unlike running two instances of DRBD on the same system (consensus is "it should work", "it seems to work", "not really what it was intended for", and "probably not a good idea"), this is one of the intended use cases for this feature of the Linux MD driver. I think it would be wise to investigate that further before pursuing DRBD for this use. https://raid.wiki.kernel.org/index.php/Write-intent_bitmap -- Nelson Hicks