Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Tuesday 05 October 2004 17:13, Benoit.Ropartz at alcatel.fr wrote:
> Ok i have tested writes speed during normal operation and i got sames
> results
> with two versions ~ 600 Ko/s.
>
> But why i find difference when syncing is active (30ko=>2ko) ???
Just read the changelog of 0.7.5, the write speed barrier of 30MB/s has gone
with this version. Probably you are now syncing with the full speed of your
hardware (check the logs or /proc/drbd during the synchronisation to get the
sync speed) and with the previous version only at 30MB/s and so there was l
more capacity left for normal writing operation.
>
> And 2ko/s is very small....
Btw, what does the unit 'ko' mean? kilo operations?
>
> Your idea is fine (rate=5M) if there is only one resource synchronised but
> not if there are many.
Sorry, I don't understand what you mean. The syncer {} definition is per drbd
group. Anyway, same hardware devices should be different sync groups, e.g.
resource resource drbd0 {
...
syncer {
rate 5M;
group 1;
}
...
}
resource resource drbd1 {
...
syncer {
rate 5M;
group 2;
}
...
}
Cheers,
Bernd