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, Mar 09, 2009 at 04:17:18PM -0400, Little, Kevin wrote: > Hi, all. > > I'm using 8.0.3 DRBD on a PPC platform, MontaVista 2.6.21_mvlcge500 base (ppc64). don't. get montavista to finally acknowledge that drbd 8.0.3 is broken, and that they NEED. TO. UPGRADE. AT. LEAST. TO. LATEST. 8.0. NOW. this has been discussed on the list a few times, and even in private mail with some of the montavista folk. but they apparently are either overstrained or completely resistent to upstream counsel. not meaning to offend anyone, but really, have a look at the drbd changelog of the last two years (8.0.3 was May 2007!). > I'm running into the "attempt to access beyond end of device" > condition written up on the list, this indeed has nothing to do with drbd version, but with operator error. > but it looks to me that I have over > 500MB (26%) of the drbd partition still available, even after the copy > would be complete. > > I did the check of IMD space requirements as per the formula given by > Lars, and it looks ok (see below). > > Any insights? > Version: 8.0.3 (api:86) > SVN Revision: 2881 build by build at node-44, 2008-04-03 21:30:55 > > ## we're going to attempt to copy a 311MB file... > > primarybox:/public/klittle# ls -lh /sw/bigfile > -rwxr-xr-x 1 root root 311M Mar 9 2009 /sw/bigfile > > ## target is /public/klittle; /public is mounted on device /dev/drbd3: > > primarybox:/public/klittle# df . > Filesystem 1K-blocks Used Available Use% Mounted on > /dev/drbd3 1929068 1019452 811624 56% /public > > ## thus a copy of bigfile will take ~304000 1k-blocks. > ## this should leave ~507600 blocks (26%) available after the copy is done. > > ## /public was a existing partition prior to being put under drbd; > ## it was mounted on device /dev/sda7 > > primarybox:/public/klittle# grep -e drbd3 -e sda7 /proc/partitions > 8 7 1959898 sda7 > 147 3 1959800 drbd3 > ## check IMD space requirements: > ## 1959898/(4*8*512*2)= 59.81134033203125 > ## 60 + 36= 96 > ## 1959898-96= 1959802 check! cool. unfortunately off track. you mkfs'ed on sda7, not on drbd3. so your file system _thinks_ it has the size of sda7 available, and feels free to allocate space somewhere beyond the end of drbd3. which fails. so don't do that. you apparently have found the other threads about this, so you already know how to recover. > ## note: for logistical/development reasons, which would that be? > I have skipped /dev/drbd1: > primarybox:/public/klittle# ls -la /dev/drbd* > brw-r--r-- 1 root root 147, 0 Jan 1 12:00 /dev/drbd that should be drbd0, not just "drbd". > brw-r--r-- 1 root root 147, 2 Jan 1 12:00 /dev/drbd2 > brw-r--r-- 1 root root 147, 3 Jan 1 12:00 /dev/drbd3 > > > ## but, even with head-room of 500MB (26%)... -- : Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com DRBD® and LINBIT® are registered trademarks of LINBIT, Austria. __ please don't Cc me, but send to list -- I'm subscribed