We are attempting to perform a point in time restore for one of several
thousand users on a large mail server. The file system involved contains
almost four million files. The server is at level 4.2.3.2 and running
under OS/390. The client is at level 5.1.1.0 and running under Linux
at a 2.4.9-34s kernel level. The system administrator has declined to
offer any estimate of the number of files to be restored. Several
attempts to carry out the restore have been terminated to allow normal
backup processing to run. In desparation we have suspended normal
expiration and reclamation processing to enable the restore to get as
much of the server's resources as possible. We also restarted the
server to ensure that server memory management was as clean as possible.
The restore has been running for nearly an hour with no data movement to
or from the client and no tape mount requests. It is presumably figuring
out which files to restore and where the files are located. Tivoli
has never thought it necessary to provide any sort of progress indicator
for this kind of processing. However, 'query db f=d' commands show
almost eight million buffer requests, most of them presumably from the
restore. How does the number of buffer requests scale with the number
of files to be restored?