[DRBD-user] slow performance copy file on drbd resource

[morbus2000 at web.de]

hi,

Hardware / Software: 
two Supermicro Server 8GB RAM, 2xQuadcore Xeon 5x
3ware 9550 Raid 5 with 6*320GB Seagate, 
Network for DRBD between the two systems is a direct 10 GB connection (Intel 82598EB 10GbE AF Network Adapter) only for DRBD Sync
SLES 10 SP2, 2.6.16.60-0.39.3-xen #1 SMP x86_64
DRDB 8.2.6
Raid 5 > phy. Disk > LVM > DBBD > LVM > XEN > VBD
one lv on top of DRBD as OCFS2 for the DomUŽs

... so we test the following

set
# blockdev --setra 8 /dev/sda (default)
# echo 512 > /sys//block/sda/queue/nr_requests
# echo deadline > /sys/block/sda/queue/scheduler
# echo 20 > /proc/sys/vm/dirty_background_ratio
# echo 60 > /proc/sys/vm/dirty_ratio

copy 2,7GB iso file from /tmp > lv on top of drbd /dev/vgmirror/test - time 7:43 min

then we set read-ahead
# blockdev --setra 16384 /dev/sda (only change)
# echo 512 > /sys//block/sda/queue/nr_requests
# echo deadline > /sys/block/sda/queue/scheduler
# echo 20 > /proc/sys/vm/dirty_background_ratio
# echo 60 > /proc/sys/vm/dirty_ratio

copy 2,7GB iso file from /tmp > lv on top of drbd /dev/vgmirror/test - time 0:35 min

now we copy the same file direct on top of drbd - time 3:42 min

then we attach the same phy lv /dev/vgmirror/test to DomU
# xm block-attach 10 phy:/dev/vgmirror/test /dev/xvdb w

copy the *.iso file from Host /tmp > DomU:/mnt which mount to /dev/vgmirror/test
# scp /tmp/*.iso domU:/mnt/
after approximately copy 1GB eth transfer rate goes bck to 10 Mb/s
after approximately copy 2GB -- stalled-- (time 5:09 min)

I don t understand this behavior?

my config

/etc/drbd.conf
-----------------
global {
 usage-count yes;
 # minor_count 5;
 # dialog-refresh 5;
 disable-ip-verification;
}

common {
 syncer { rate 600M; al-extents 257; }
}

resource drbd0 {
 protocol       C;

 handlers {
  split-brain "/usr/lib/drbd/notify-split.brain.sh root";
          }

 startup {
  wfc-timeout 0;
  degr-wfc-timeout 120;
  become-primary-on both;
         }

  disk {
   on-io-error detach;
   max-bio-bvecs 1;
      }

  net {
   allow-two-primaries;
   after-sb-0pri discard-zero-changes;
   after-sb-1pri discard-secondary;
   after-sb-2pri disconnect;
      }

 on enterprise5 {
  device        /dev/drbd0;
  disk          /dev/vglocal/lvlocal;
  address       192.168.1.15:7788;
  flexible-meta-disk    internal;
 }

 on enterprise6 {
  device        /dev/drbd0;
  disk          /dev/vglocal/lvlocal;
  address       192.168.1.16:7788;
  flexible-meta-disk    internal;
 }
}

Why is write performance so low respectively "stalled" ? 
What else can I do to get better write performance?

help me, please.

Regards
Andreas

________________________________________________________________
Neu: WEB.DE Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate
für nur 19,99 Euro/mtl.!* http://produkte.web.de/go/02/