DRBD Operations With Many Files Slow

Question

I need to set up file replication between two geographically different DC. My first idea was with GlusterFS, but it doesn't work well with high latency Blicks, and its geo replication is not bidirectional.

Set up DRBD+OCFS2 and it is replication however when you have operations with multiple files the replication is very slow.

I ran some test with DD

root@nfs02:/gluster/r0# dd if=/dev/zero of=./testdd bs=1M count=500
500+0 records in
500+0 records out
524288000 bytes (524 MB, 500 MiB) copied, 35.5622 s, 14.7 MB/s

When doing with DD the speed on the network interface is equal to the speed of DD using ftop.

But when I make a cp or chmod with several files the traffic is at the kb speed and is very slow.

My config DRBD

resource r0 {
        protocol C;
        meta-disk internal;
        device /dev/drbd0;
        disk /dev/nfs/nfs;

        startup {
                wfc-timeout 20;
                become-primary-on both;
        }

        net {
                allow-two-primaries yes;
                after-sb-0pri discard-zero-changes;
                after-sb-1pri discard-secondary;
                after-sb-2pri disconnect;
                rr-conflict disconnect;
                rcvbuf-size 10M;
                sndbuf-size 10M;
                no-tcp-cork;
                unplug-watermark 16;
                max-buffers 40k;
                max-epoch-size 8000;
        }

        disk {
                on-io-error detach;
                resync-rate 250M; 
                no-disk-flushes ;
                no-disk-barrier;
                no-md-flushes;
                c-plan-ahead 0;
                c-fill-target 1M;
                c-min-rate 250M;
                c-max-rate 800M;
        }

        syncer {
               rate 90M;
        }

        on nfs01 {
                address  172.16.100.101:7789;
        }

        on nfs02 {
                address  172.16.200.101:7789;
        }
}

Thanks!

DRBD Operations With Many Files Slow

0 Answers0