Hi,
I haven't been able to reproduce this locally. Can you generate some
additional debug output on the osd by starting with
./cosd /dev/md0 -d --debug_osd 20 --debug_ebofs 20
and send me the contents of out/osd0 (or whichever osd sees the csum
errors)?
thanks-
sage
On Tue, 23 Sep 2008, el
I am evaluating ceph on a cluster of three machines. Each time I try to
copy a 4.6G file using the kernel client (2.6.24.4) all ceph data
becomes corrupted.
The following messages are found in the log:
080923.115611.857658 ebofs.oc.rx_finish bad csum wanted f8692852eac0562d
got 0 for object bloc
