I had the same question several months ago. Here's an explanation of the settings that canbe found in the tsmjbbd.ini file in the TSM install directory. Basically revolves around PreserveDBOnExit and deferFSMonStart values.
NotifyBufferSize default value is 0x00100000 (1 megabyte). For
example:
NotifyBufferSize=0x00200000
..
From what I can see on my system is if the above memory is avaialbe, the
buffer will use up to this size.
..
PreserveDbOnExit Setting:
..
This setting allows a journal to remain valid when a journaled file
system goes offline and comes back online. This is useful for preserving
the journal during system reboots, cluster failovers, and resource
movement.
File systems go offline when the journal service stops or when the file
system is removed from the configuration file. File systems come back
online when the journal service is started or when the file system is
added to the configuration file.
This setting allows a journal based backup to continue processing when
the service is restarted (or the file system comes back online) without
performing a full incremental backup.
..
Note: Any change activity which occurs while the journal service is not
(or the file system is offline) will not be recorded in the journal.
..
In a clustered environment, shared resources can move to different
machines in the cluster. The journal service running on each machine in
the cluster must include these shared resources in the list of journaled
file systems. The journal service running on the machine which currently
owns the resource will actively be journaling the shared resource while
other journal services on machines in the cluster which do not own the
resource must defer journaling until the resource becomes available (or
is moved to that machine). The configuration settings deferFSMonStart,
deferRetryInterval, and logFSErrors allows deferment for a file
system until the file system is available and accessible.
A value of 1 indicates that the journaled file system journal database
will not be deleted when the journal file system goes offline. The
database will also be valid when the journal file system comes back
online. This value should be used with caution because any file system
change activity which occurs while the journaled file system is offline
will not be reflected in the journal database.
An example for not deleting the journal database upon exit is:
[JournaledFileSystemSettings.D:\]
; Do not delete the journal when D:\ goes offline
;
PreserveDbOnExit=1
An example to defer journaling until the file system journal directories
are valid is:
[JournalSettings]
Place journal files in directory on each journaled file system
;
journalDir=\tsmjournal
[JournaledFileSystemSettings]
;
;journal c:, d:, and f:
;
JournaledFileSystems=c: d: f:
Override stanza to defer starting journaling for f:\
; until it is a valid file system
;
[JournalFileSystemSettings.f:\]
Keep database valid if file system goes offline
;
PreserveDBOnExit=1
Defer journaling until file system and journal directory
; are valid
;
deferFSMonStart=1
Attempt to start journaling every 120 seconds when deferred
;
deferRetryInterval=120
..
deferFSMonStart Setting:
..
This setting defers starting a file system monitor thread in the
following cases:
v When the specified journaled file system is not valid or available
v The journal directory for the specified journaled file system cannot
be accessed or created
The file system monitor thread will attempt to restart at the interval
specified by the deferRetryInterval setting until the thread
successfully starts.
A value of 1 indicates that the setting is on. A value of 0 indicates
that the setting is off. The default value is off (set to 0) .
..
deferRetryInterval Setting:
..
This setting indicates the value in seconds that a deferred file system
monitor thread will continue to attempt to restart until the thread is
successfully started. The default value is 60 seconds.
..
logFSErrors Setting:
..
This setting Indicates if errors encountered accessing a journaled file
system or journal directory should be logged. Errors are logged both to
the jbberror.log and the NT event log.
If this setting is on and deferFSMon is set, the first error which
causes the file system monitor thread to be deferred will be logged.
Subsequent errors will not be logged.
A value of 1 indicates that the setting is on. A value of 0 indicates
that the setting is off.
Regards, M