Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Mon, Oct 01, 2012 at 04:04:47PM -0600, randal k wrote: > DRBD Collective! > We are planning on expanding an existing DRBD resource, and have a pretty > straightforward question on how to do it. I have read every thread, > document and more, and it really does boil down to this: does a half-way > decent RAID controller qualify as a dynamic backing block device? Of course ;-) > Here is our pancake, which isn't too complex: > > [xen] > [vm filesystem] > LVM [because xen wants it] > iSCSI > DRBD (on a block device /dev/sdb, not a partition or LVM) > LSI 9260i RAID10 > 8x HDDs > > We would like to make 8x HDDs --> 12x HDDs. We know that we can use the > various LSI tools to expand the 8-way raid10 to 12 way, and this is very > straightforward. I also know that we can do a rescan of the /dev/sdb device > so that Linux can see the increased size. We do that on both > primary/secondary, both of the servers think they now have the increased > size, and then the DRBD resize. > > Once the array is expanded to 12 drives, can we do the online drbdadm > resize? The backing block device would already be bigger, so we *think* > that should work, but we are unsure. > > Any help is very, very appreciated. Thank you in advance! Anything more recent than 8.3.7 "remembers" the last known device size, so even if the device has been grown offline, and the drbd internal meta data is no longer to be found at the (new) end of the device, we can look for it at the last known position. If we can find it there, we move it to where it now needs to be, and that's it. So yes, do what you have to do, and as long as the existing data stays in place, this is supposed to "just work". You still may want to do this in a "quiet" period anyways, just in case ;-) We may need to update the "howto resize" section. -- : Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com