[DRBD-user] high i/o wait on primary server when reading
crsurf at terra.com.br
Thu Sep 23 17:23:10 CEST 2004
I´m having high i/o wainting (100%) on primary server when only reading data on primary server, why this occur? Theoretically ready operations are not replicated, so why occur this delay?
I observed in /proc/drbd of primary server that are a pending block in most part of time
4: cs:Connected st:Primary/Secondary ns:11681676 nr:0 dw:180576 dr:18115376 pe:1 ua:0
And in secondary server are a unacknowledged block in most part of time
4: cs:Connected st:Secondary/Primary ns:0 nr:11676268 dw:11676268 dr:0 pe:0 ua:1
This is normal?
The database block size is 2k, how can I adjust drbd to work with the same size of database block?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the drbd-user