Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
I have Debian 5.0.8 with proxmox kernel 2.6.35.
My DRBD.conf:
global { usage-count no; }
common { syncer { rate 60M; } }
resource r0 {
protocol A;
device /dev/drbd0;
disk /dev/sda1;
meta-disk internal;
on srvkvm8core {
address 10.0.0.1:7788;
}
on srvkvm4core {
address 10.0.0.2:7788;
}
disk {
no-disk-flushes;
no-md-flushes;
no-disk-barrier;
}
}
With this config, I do:
drbdadm create-md r0
drbdadm up r0
drbdadm -- --overwrite-data-of-peer primary r0
dd if=/dev/zero of=/dev/drbd0 bs=1G count=4 conv=fdatasync
And i get:
4294967296 bytes (4.3 GB) copied, 120.165 s, 35.7 MB/s
sda1 is a partition on a hard drive, write speed of which is 70+MB/s, so I
get less then 50% performance. As you can see I've added no-disk-... options
but they do not produce performance change. What am I doing wrong? Or maybe
drbd is just slow by design while there is no replication? (maybe it
fiercely tries to contact peer etc which lowers speed, and if I give it one,
it will work fine?)
I don't have another server right now, but I will try to make a VM as a peer
on the same host, and passthrough another hdd into it, make drbd in VM and
test, but I'm afraid that will not make difference.
--
View this message in context: http://old.nabble.com/Initial-offline-DRBD-setup---slow-write-speed.-How-to-improve--tp30976028p30976028.html
Sent from the DRBD - User mailing list archive at Nabble.com.