[DRBD-user] DRBD resync QoS

[wang xuchen]

Hi All,

Here is one problem I have encountered but haven`t figured out the answer
yet.

In my testing environment, I have ten DRBD devices configured as
primary/primary. I do have a cluster management software that
is irrelevant to the topic. My client is running IOMeter with queue depth
set to 16 for each target.

If I reboot one of server, all IOs will failover to its partner, no error is
reported. However, during DRBD resync, it seems to me that DRBD dedicates
all its resources to resync traffic which leaves new IOs from client pending
too long and eventually timed out. Does DRBD have a mechanism that can tune
the aggressiveness of the the resync operation? In other word, how can I
tell DRBD to favor application pending(ap) IO as opposed to resync IO so
that from client`s point of view QoS is guaranteed.

resource drbd10 {
  on FA33 {
    device /dev/drbd10;
    disk /dev/disk/by-id/scsi-360030480003ae2e0159207cc2a2ac9d4;
    address 192.168.251.1:7799;
    meta-disk internal;
  }
  on FA34 {
    device /dev/drbd10;
    disk /dev/disk/by-id/scsi-360030480003ae32015920a821ca7f075;
    address 192.168.251.2:7799;
    meta-disk internal;
  }
  net {
    allow-two-primaries;
    after-sb-0pri discard-younger-primary;
    after-sb-1pri discard-secondary;
    after-sb-2pri violently-as0p;
    rr-conflict violently;
    max-buffers 8000;
    max-epoch-size 8000;
    unplug-watermark 16;
    sndbuf-size 0;
  }
  syncer {
    rate 300M;
    verify-alg crc32c;
    al-extents 3800;
  }
  startup {
    become-primary-on both;
  }
  handlers {
    before-resync-target "/sbin/before_resync_target.sh";
    after-resync-target "/sbin/after_resync_target.sh";
  }
}

Any help is appreciated.
Thanks
Ben
Commit yourself to constant self-improvement
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20110621/476028ea/attachment.htm>