dbarker at visioncomm
Jul 22, 2010, 9:20 AM
Post #2 of 5
(604 views)
Permalink
|
|
Re: STRANGE ISSUE - disk partition deleted after resync
[In reply to]
|
|
I'm not sure why you wanted to "erase" /dev/sdb1, but to do so, you'd use dd
with an outfile of /dev/sdb1, not /dev/sdb. You cleared the first megabyte
of the disk, and that includes the partition table. Maybe you wanted to
create the metadata and it said metadata was already there. So, use drbdadm
-- --force create-md <res>. dd not needed and you can't specify the
incorrect device. If the drbd configuration specification is correct, the
proper device will get the metadata.
If the disk does not contain data, there is no point in synchronizing it.
Just mark it empty and connect (drbdadm -- --clear-bitmap new-current-uuid
<resource>). It's instant vs 30 minutes. Details below.
Thirdly, if the disk is dedicated to drbd, why a partition at all, just use
the whole disk, (/dev/sdb vs /dev/sdb1). You just waste space for the
partition table and confuse the issue.
Dan in Atlanta
From: drbd-user-bounces [at] lists
[mailto:drbd-user-bounces [at] lists] On Behalf Of Roberto Fastec
Sent: Thursday, July 22, 2010 12:04 PM
To: drbd-user [at] lists
Subject: [DRBD-user] STRANGE ISSUE - disk partition deleted after resync
I'm doing my first tests with drbd.
My configuration is two pcs with two hdd each.
One drive is xenserver the second is drbd dedicated. On the drbd dedicated
drive, I created /dev/sdb1 which in drbd.conf is assigned to drbd0.
Well, because it is a test environment, I messed up the things a bit and so
I had to erase drbd0 (which still is not clear to me which is the correct
procedure, but I did it) and, to be sure, I also deleted and recreated
/dev/sdb1.
The idea was to start over with the creation of drbd0 resource.
The first issue I had was about one error that was exiting the command
drbdadm create-md drbd0. Googleing I've found the solution was to (I'm not
so strong with dd) issue this command dd if=/dev/zero bs=1M count=1
of=/dev/sdb; sync, looks like that first 1MB data are moved (?), anyway this
worked.
Then drbdadm create-md drbd0 worked again and I finished the sequence
# drbdadm create-md drbd0 #Create device metadata
# drbdadm attach drbd0 #Attach to backing device
# drbdadm syncer drbd0 #Set synchronization parameters
# drbdadm connect drbd0 #Connect to peer
last I launched the primary command and started the sync
# drbdadm -- --overwrite-data-of-peer primary drbd0
and the sync worked fine, with a final average speed of 81MB/sec, in 35
minutes the 200GB volume was synced, I also copied and pasted to my notes
the cat /proc/drbd command result while syncing
[root [at] xenserver- dev]# cat /proc/drbd
version: 8.3.8.1 (api:88/proto:86-94)
GIT-hash: 0d8589fcc32c874df57c930ca1691399b55ec893 build by
root [at] localhost, 2010-07-17 10:04:02
0: cs:SyncSource ro:Primary/Secondary ds:UpToDate/Inconsistent C r----
ns:935040 nr:0 dw:0 dr:11435392 al:0 bm:697 lo:145 pe:0 ua:145 ap:0 ep:1
wo:b oos:183876844
[>...................] sync'ed: 5.9% (179564/190732)M
finish: 0:34:18 speed: 89,308 (76,232) K/sec
when finished, it was late night and so I turned off the secondary, last the
primary and I went at home.
Now I'm here to go on with my tests. Turned on the primary, turned on the
secondary and with my big surprise, I've found this
[root [at] xenserver- ~]# drbd-overview
0:drbd0 Unconfigured . . . .
so I also lauched cat /proc/drbd
[root [at] xenserver- ~]# cat /proc/drbd
version: 8.3.8.1 (api:88/proto:86-94)
GIT-hash: 0d8589fcc32c874df57c930ca1691399b55ec893 build by
root [at] localhost, 2010-07-17 10:04:02
0: cs:Unconfigured
so I issued
[root [at] xenserver- ~]# drbdadm up drbd0
Can not open device '/dev/sdb1': No such file or directory
Command 'drbdsetup 0 disk /dev/sdb1 /dev/sdb1 internal --set-defaults
--create-device --max-bio-bvecs=1 --on-io-error=detach' terminated with exit
code 20
drbdadm attach drbd0: exited with code 20
and because of the message, I checked with fdisk -l
[root [at] xenserver- ~]# fdisk -l
Disk /dev/sda: 500.1 GB, 500107862016 bytes
255 heads, 63 sectors/track, 60801 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/sda1 * 1 523 4194304 83 Linux
Partition 1 does not end on cylinder boundary.
/dev/sda2 523 1045 4194304 83 Linux
/dev/sda3 1045 60801 479995393 8e Linux LVM
Disk /dev/sdb: 1000.2 GB, 1000215724032 bytes
255 heads, 63 sectors/track, 121602 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk /dev/sdb doesn't contain a valid partition table
Yes it is, /dev/sdb1 is disappeared... on both computers...
Below you find the drbd.conf.
My question are:
- which is the correct mode to erase one drbd resource ?
- is it possible what described above?
- to be able to do such "disaster" (I'm thinking about a production
environment), what could be the error/mistake?
- with the following configuration file, both the drbd are starting in
secondary mode, maybe this is wanted by drbd concept, but I can't understand
it very well.
I mean, when everything works fine, I have one primary and one secondary.
If I do a clean shutdown and I shut down first of all the secondary, next I
shut down the primary, why starting up first the primary and last the
secondary both of them are secondary? Is it the only solution
become-primary-on server-1 statement? If yes... I have another issue.. this
didn't worked... and powercycling the two pcs in the correct order, again
returned a secondary/secondary situation.
- Is it correct to use the server name in the following statement
become-primary-on server-1 ?
- Which are the downsides of usignthis setting in the drbd.conf:
become-primary-on server-1 ?
Thank you for any tip and help, follows the conf file
Robert
drbd.conf
# You can find an example in /usr/share/doc/drbd.../drbd.conf.example
#include "drbd.d/global_common.conf";
#include "drbd.d/*.res";
global {
usage-count yes;
}
common {
syncer {
rate 1G;
verify-alg md5;
csums-alg md5;
}
}
resource drbd0 {
protocol C;
startup {
#become-primary-on xenserver-2;
}
net {
cram-hmac-alg md5;
shared-secret "ColdWater";
sndbuf-size 0;
rcvbuf-size 0;
data-integrity-alg md5;
}
disk {
max-bio-bvecs 1;
on-io-error detach;
}
on server-1 {
device /dev/drbd0;
disk /dev/sdb1;
address 10.1.1.2:7789;
meta-disk internal;
}
on server-2 {
device /dev/drbd0;
disk /dev/sdb1;
address 10.1.1.3:7789;
meta-disk internal;
}
handlers {
split-brain "/usr/lib/drbd/notify-split-brain.sh root";
}
}
|
