Hi Rick,
originating host generate flows to all nodes where replica will be
stored, e.g. if you've configured three replicas, then:
- in general, at least two copies of data will pass the NIC
- but if you're using volume on 'diskless' host, three copies of data
will pass the NIC
similar arithmetic apply to two replicas.
Please see my recent message with subject 'drbd & multicast'.
On 03.04.2020 18:55, Rick Gutierrez wrote:
hi list, i am installing drbd 8.4.x in centos 7 in a couple of boxes,
i need to replicate my mail server, but i am having speed problems in
the replication, i have tried different options within the
configuration and without success.
I have my network at 1gb ethernet, my switch is cisco 2960-S, I am
replicating 1.3TB on 7200RPM disks, but the speed is terrible, very
slow.
version of rpm
drbd84-utils-9.6.0-1.el7.elrepo.x86_64
kmod-drbd84-8.4.11-1.1.el7_6.elrepo.x86_64
drbd84-utils-sysvinit-9.6.0-1.el7.elrepo.x86_64
my setting:
disk {
c-plan-ahead 0;
on-io-error detach;
no-disk-flushes ;
no-disk-barrier;
c-min-rate 90M;
c-max-rate 120M;
net {
allow-two-primaries;
cram-hmac-alg sha1;
shared-secret "C2qz2wsx";
max-buffers 36k;
sndbuf-size 1024k ;
rcvbuf-size 2048k;
}
any idea?
--
Volodymyr Litovka
"Vision without Execution is Hallucination." -- Thomas Edison
_______________________________________________
Star us on GITHUB: https://github.com/LINBIT
drbd-user mailing list
[email protected]
https://lists.linbit.com/mailman/listinfo/drbd-user
