Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On 01/27/2011 11:13 AM, Nauman Yousuf wrote:
> here
>
> Primary DRBD
>
> skip {
> Lustre storage cluster replication setup
> LUSTRE OSS active/active Replication
>
> }
>
> global {
> minor-count 2;
> }
>
> resource r0 {
> protocol C;
>
> startup {
> degr-wfc-timeout 120;
> }
>
> disk {
> on-io-error detach;
> }
>
> net {
> }
>
> syncer {
> rate 30M;
> group 1;
> al-extents 257;
> }
>
> on typhoon.example.com {
> device /dev/drbd0;
> disk /dev/cciss/c0d0p8;
> address 10.65.200.21:7788;
> meta-disk /dev/cciss/c0d0p6[0];
> }
>
> on cyclops.example.com{
> device /dev/drbd0;
> disk /dev/cciss/c0d0p8;
> address 10.65.200.30:7788;
> meta-disk /dev/cciss/c0d0p6[0];
> }
> }
I'm not familiar with Lustre operation. Does /dev/cciss/c0d0p8 mean that
both nodes in fact use the same chunk of Lustre data? Because that would
probably be bad.
> resource r1 {
> protocol C;
>
> startup {
> degr-wfc-timeout 120;
> }
>
> disk {
> on-io-error detach;
> }
>
> net {
> }
>
> syncer {
> rate 30M;
> group 1;
> al-extents 257;
> }
>
> on typhoon.example.com {
> device /dev/drbd1;
> disk /dev/cciss/c0d0p9;
> address 10.65.200.21:7789;
> meta-disk /dev/cciss/c0d0p7[0];
> }
>
> on cyclops.example.com {
> device /dev/drbd1;
> disk /dev/cciss/c0d0p9;
> address 10.65.200.30:7789;
> meta-disk /dev/cciss/c0d0p7[0];
> }
> }
>
>
> Secondary DRBD
>
> skip {
> Lustre storage cluster replication setup
> LUSTRE OSS active/active Replication
>
> }
>
> global {
> minor-count 2;
> }
>
> resource r0 {
> protocol C;
>
> startup {
> degr-wfc-timeout 120;
> }
>
> disk {
> on-io-error detach;
> }
>
> net {
> }
>
> syncer {
> rate 30M;
> group 1;
> al-extents 257;
> }
>
> on typhoon.example.com {
> device /dev/drbd0;
> disk /dev/cciss/c0d0p8;
> address 10.65.200.21:7788;
> meta-disk /dev/cciss/c0d0p6[0];
> }
>
> on cyclops.example.com {
> device /dev/drbd0;
> disk /dev/cciss/c0d0p8;
> address 10.65.200.30:7788;
> meta-disk /dev/cciss/c0d0p6[0];
> }
> }
>
> resource r1 {
> protocol C;
>
> startup {
> degr-wfc-timeout 120;
> }
>
> disk {
> on-io-error detach;
> }
>
> net {
> }
>
> syncer {
> rate 30M;
> group 1;
> al-extents 257;
> }
>
> on typhoon.example.com {
> device /dev/drbd1;
> disk /dev/cciss/c0d0p9;
> address 10.65.200.21:7789 <http://10.65.200.21:7789>;
> meta-disk /dev/cciss/c0d0p7[0];
> }
>
> on cyclops.example.com {
> device /dev/drbd1;
> disk /dev/cciss/c0d0p9;
> address 10.65.200.30:7789;
> meta-disk /dev/cciss/c0d0p7[0];
> }
> }
>
> I am using DRBD to replicate Lsuter data .
How? Where is your MDS? What are your OSSs? Where are the nodes that use
DRBD and how do they access Lustre? (Sorry, I cannot infer that from the
/dev/cciss/... stuff).
Also, please be careful to CC the list.
Regards,
Felix