You're going to have to do more research on your operating system's
implementation of NFS.
rsize,wsize control the maximum (negotiated) buffer size between the
filer and the OS.
nointr - do NOT allow file operations to be interrupted
hard - send console warning if device has "major timeout" and continue
trying the operation.
tcp - use tcp protocol (traffic is acknowledged) instead of the UDP
(send and assume receipt) protocol
vers - identifies the protocol the computer will use to establish
communication with the server.
None of these parameters specifically relate to DB2. The questions you
need to ask are:
1. When does the operation system return a "successful write" response
to the application for NFS mounted volumes?
2. How does buffering effect the response above?
Don't forget that DB2 writes logs at commit time and may not write data
until a later time when a page cleaner returns it to disk. Recovery from
a backup and the logs wouldn't care if the last n data writes failed due
to an nfs buffering issue.
Phil Sherman
lltong wrote:
Hi,
Does anybody have any experience with DB2 on NetApp Filer using NFS? I
am reading the documentation that it suggests to mount the data volumn
to: [StorageSystemName]:[VolName] [MountPoint] nfs hard,
rw,nointr,rsize=32768,wsize=32768,bg,vers=3,tcp 0 0
So my question is: in case of server crash, do we lose the data up to
the NFS buffered size? Is there any option that we can get away with
the potential data loss?
Thanks,
Lisa
