Hi all,
Finally, I got drbd 0.6.1pre2 running on my two SMP-Boxes.
Just want to tell you about the StandAlone thing:
It happens when I put a statement "disk-size=400000" (for example) in
drbd.conf.
Then I start drbd on one node, anwering yes to make it primary. cat
/proc/drbd tells me it is connected primary/unknown
Starting drbd on the other node makes both nodes StandAlone and disk is
beeing accessed as if they sync or check something. All counters show zeros.
This state lasts obviously a long time, I don´t know wether they will
connect some time ´cause I didn´t wait for that, always aborted it after 2
hours. The underlaying Partition is 512MB of size (for testing puroses.)
Removing the disk-size entry, everything works fine and both nodes connect.
I´m impressed by the much better performance of drbd now. Another Partition
of mine ist 4GB and it does a syncall in about 15 minutes. It´s on a
scsi-disk, Fujitsu 9,1GB - 7200rpm - not the fastest one I think. Maybe that
is why it has no effect on the time if I use 100Mbit/s Fast Ethernet or
1000Mbit/s Gigabit Ethernet for drbd - the time is equal. Or, my
configuration isn´t optimized. In drbd.conf I have sync-rate=30000 and
tl-size=1024. Too high? Top sais that CPU-Time for drbd is about 4% when
syncing. Maybe I should try faster disks.
My Hardware is: 2 FSC Primergy E200 with each 2*866 PIII, ServerWorksLE,
256MB reg.SDRAM, NCR UW-SCSI160 onBoard, SCA-Backplane, Intel Pro/100
Serveradapter Onboard and Intel 1000T 64Bit PCI Adapter with Crossover-Link
between.
Sincerely
Heino Schneider
Finally, I got drbd 0.6.1pre2 running on my two SMP-Boxes.
Just want to tell you about the StandAlone thing:
It happens when I put a statement "disk-size=400000" (for example) in
drbd.conf.
Then I start drbd on one node, anwering yes to make it primary. cat
/proc/drbd tells me it is connected primary/unknown
Starting drbd on the other node makes both nodes StandAlone and disk is
beeing accessed as if they sync or check something. All counters show zeros.
This state lasts obviously a long time, I don´t know wether they will
connect some time ´cause I didn´t wait for that, always aborted it after 2
hours. The underlaying Partition is 512MB of size (for testing puroses.)
Removing the disk-size entry, everything works fine and both nodes connect.
I´m impressed by the much better performance of drbd now. Another Partition
of mine ist 4GB and it does a syncall in about 15 minutes. It´s on a
scsi-disk, Fujitsu 9,1GB - 7200rpm - not the fastest one I think. Maybe that
is why it has no effect on the time if I use 100Mbit/s Fast Ethernet or
1000Mbit/s Gigabit Ethernet for drbd - the time is equal. Or, my
configuration isn´t optimized. In drbd.conf I have sync-rate=30000 and
tl-size=1024. Too high? Top sais that CPU-Time for drbd is about 4% when
syncing. Maybe I should try faster disks.
My Hardware is: 2 FSC Primergy E200 with each 2*866 PIII, ServerWorksLE,
256MB reg.SDRAM, NCR UW-SCSI160 onBoard, SCA-Backplane, Intel Pro/100
Serveradapter Onboard and Intel 1000T 64Bit PCI Adapter with Crossover-Link
between.
Sincerely
Heino Schneider