19 May
2000
19 May
'00
8:13 p.m.
The CP to CP command is:
wafl_susp -w
Look at cp_from_log_full and cp_from_cp. I think they mean:
How many times the checkpoint happened (before 10 sec. delay) because the
log in
NVRAM was full (this can mean the filer writes too much, but don't worry TOO much - it can be
even
100,000).
The 2nd one I cannot remember... :(
cp_from_cp means that the NVRAM filled up so fast, it couldn't complete the last write before the next one needed to be completed. That's when your filer *really* starts to slow down, because your client is waiting on the disk drives.
Bruce