Hi all.
This is a late summary and an incomplete one, sorry about that.
The machine crashed for the second time with "kernel panic: log_write,
log half full". Then our users *managed to find time for me to do a
patch* :-)
As I was informed, this is a known bug with AdvFS which is only
partially fixed by the last, SP8 for Digital UNIX 4.0D. One thing that
might help remedy the situation is to increase number of log pages on
the domain. I will try that as a step 2.
The problem is usr_domain holds both usr and oracle file-sets, 12 GB of
data and DATs are extremly unreliable. Oh well, I still have 6x4 GB
disks, which I can plug into the KZPAC :-) and use them as a backup
space.
Some time ago I saw an undocumented way of increasing logpage number,
with "switchlog -l <num>" and two volumes. I tried "strings
/sbin/advfs/switchlog | grep l" it shows no parameters. ???
Nix.
Received on Tue Jan 23 2001 - 07:00:13 NZDT