Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Dear list,
I'm trying to build an Lustre file system with some fail-over nodes. I
install drbd but realize it take very long time to do init sync.
Here is sum of my config:
CentOS5/drbd82
1 Gig-E, straight through cable connect via switch (iperf report 902MB/s)
(I've tried straight cable direct connect but it seems slower)
I/O Scheduler: headline
sync rate set to:60MB
And here what I got:
[root at MGS2 ~]# cat /proc/drbd
version: 8.2.7 (api:88/proto:86-88)
GIT-hash: 61b7f4c2fc34fe3d2acf7be6bcc1fc2684708a7d build by
root at localhost.localdomain, 2008-11-19 06:52:49
0: cs:SyncTarget st:Secondary/Primary ds:Inconsistent/UpToDate C r---
ns:0 nr:6218720 dw:6210528 dr:0 al:0 bm:378 lo:258 pe:2683 ua:256 ap:0
ep:1 wo:b oos:92119808
[>...................] sync'ed: 6.4% (89960/96025)M
finish: 6:08:03 speed: 4,164 (3,912) K/sec
I tried to Google. Seems many ppl have the same problem. I tried some
solutions (run drbdadm adjust r0, tweak some config value, etc..) but the
result was not changed.
I don't sure where is bottle-neck. Have any software to benchmark harddisk?
Thanks!
-giobuon
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20081204/b2be1cda/attachment.htm>