Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi to all,
i'm making a new installation on SLES 10 x86_64 on 2 HP Proliant DL380
multiprocessor servers with 2 Ultra 320 SCSI disks of 146 Gb in raid1 on
each using Smart Array 6i hardware raid controller. Both servers have
2GB ram.
The 2 servers are connected on public network on eth0 and connected with
cross cable on eth1 dedicated for sync of drbd, both Gigabit ethernet.
When i make scp of one big file from one server to the other i reach
50Mb/sec speed. If i force the sync of the partitions to drbd with
invalidate_remote it do not go over 12 Mb/sec.
version: 0.7.21 (api:79/proto:74)
SVN Revision: 2341 build by lmb at dale, 2006-08-10 14:19:54
0: cs:SyncSource st:Primary/Secondary ld:Consistent
ns:41370104 nr:0 dw:0 dr:41371732 al:0 bm:6109 lo:0 pe:201 ua:407 ap:0
[==============>.....] sync'ed: 70.5% (16949/57349)M
finish: 0:22:15 speed: 12,968 (12,672) K/sec
It seems like it is limited to network connection of 10Mbit.
i tried with rate option to 50M,40M,30M and it is the same.
I'm using protocol C.
Partitions are as follows:
Device Boot Start End Blocks Id System
/dev/cciss/c0d0p1 * 1 13 104391 83 Linux
/dev/cciss/c0d0p2 14 536 4200997+ 82 Linux
swap / Solaris
/dev/cciss/c0d0p3 537 3147 20972857+ 83 Linux
/dev/cciss/c0d0p4 3148 17848 118085782+ f W95
Ext'd (LBA)
/dev/cciss/c0d0p5 3148 10458 58725576 83 Linux
/dev/cciss/c0d0p6 10459 17769 58725576 83 Linux
/dev/cciss/c0d0p7 17770 17848 634536 83 Linux
Partitions for the system are p1/p2/p3, partitions to be synced are p5
and p6 (now testing only with p5), i'm using the last partition p7 as
metadisk partition.
DRBD version 0.7.21 and kernel 2.6.16.27-0.9-smp x86_64
If you have any clue on the possible cause please let me know. May be
this speed is normal for this kind of environment or may be i have to
tune some network parameters.
Thanks,
Bye,
Marcello