[DRBD-user] DRBD initial sync very slow

Lars Ellenberg lars.ellenberg at linbit.com
Fri Jun 6 22:37:28 CEST 2014

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

On Thu, May 22, 2014 at 02:48:17PM +0200, Holger Buß wrote:
> Hi,
> i had setup a Node with DRBD und Heartbeat in a Standalone Setup, because the other Node was not ready to use.
> Setup:
> - 2x 600GB Partions would be used as ressource data1 and data2.
> The Standalone Node is mounted and configured as resource data2 with ext4 filesystem. On data2 exists a lot of files. Ressource data1 was not formated and not used in DRBD Configuration yet. All works fine.
> After a few days the second Node becomes ready. The 2nd Node got also 2x 600 GB Partitions.
> First i setup the Resource data1 on the second Node as Primary. The initial setup (sync) was done in 1 hour 30 minutes with a configured rate 100M. All fine.
> cat /proc/drbd
> version: 8.3.13 (api:88/proto:86-96)
> GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by root at sighted, 2012-10-09 12:47:51
>  0: cs:SyncSource ro:Secondary/Primary ds:UpToDate/Inconsistent C r-----
>     ns:139209584 nr:9969700 dw:19939436 dr:139211637 al:0 bm:8775 lo:43 pe:36 ua:45 ap:0 ep:1 wo:b oos:481534196
>         [===>................] sync'ed: 23.4% (470248/613192)M
>         finish: 1:16:21 speed: 105,108 (101,932) K/sec
>  1: cs:Connected ro:Primary/Secondary ds:UpToDate/Diskless C r-----
>     ns:0 nr:0 dw:597294168 dr:52092731 al:222 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:629126364
> Both Nodes are directly connected only for DRBD and Heartbeat.
> On the First Node i started the DRBD sync as primary to the 2nd Node. But the initial sync from Primary (with filesystem and data on disk) was very slow.
> cat /proc/drbd
> version: 8.3.13 (api:88/proto:86-96)
> GIT-hash: 83ca112086600faacab2f157bc5a9324f7bd7f77 build by root at sighted, 2012-10-09 12:47:51
>  0: cs:Connected ro:Secondary/Primary ds:UpToDate/UpToDate C r-----
>     ns:620739192 nr:9969700 dw:19939436 dr:620741097 al:0 bm:38399 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0
>  1: cs:SyncSource ro:Primary/Secondary ds:UpToDate/Inconsistent C r-----
>     ns:2223984 nr:0 dw:598733424 dr:53982399 al:551 bm:125 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:626937968
>         [>....................] sync'ed:  0.4% (612244/614380)M
>         finish: 115:14:45 speed: 1,500 (1,480) K/sec
> The initial sync between two unformated and not mounted Partions was finished in 1h 30m, but the initial sync beween with a mounted Disk runs 120 hours?
> Is this normal?

See "DRBD and the sync rate controller, part 2"
And adjust your c-min-rate.

> Configuration (drbd.conf):
> global {
>         usage-count no;
> }
> resource data1 {
>         protocol C;
>         handlers {
>                 pri-on-incon-degr "echo o > /proc/sysrq-trigger ; halt -f";
>                 pri-lost-after-sb "echo o > /proc/sysrq-trigger ; halt -f";
>                 local-io-error "echo o > /proc/sysrq-trigger ; halt -f";

Are you really sure you mean to power off the box, hard,
if you happen to get an IO error?

Or if you happen to lose the replication link on an inconsistent Primary?

Also, you likely want to increase your al-extents...

: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

More information about the drbd-user mailing list