what to log - storage manager

[Arie Shoshani <arie@math.lbl.gov>]

We (the people working on the Storage Manager) discussed what will be
useful to log for the Storage Manager.  Here is what may be a partial
list.  All the item below log time and content.


QE LOGS:

1) Each query submitted the QE.

2) Each files-event_list that the QE submits to the QM.

3) Each msg from QE of event_list to Query Object.

4) Each msg from QE of query_estimation to Query Object.


QM LOGS:

1) When the QM recieves files-event_list message from the QE. 
This is only to verify proper communication.

2) Each file_caching request from QM to CM.

3) When the QM recieves msg from CM that a file was cached.

4) Each file_was_cached message from QM to an EI.

5) Done_with_file message from any EI to QM.


CM LOGS:

1) When the CM recieves file_caching message from the QM.
This is only to verify proper communication.

2) When a request to cache a file is made to HPSS. 

3) When the file was cached by HPSS. 

4) When the CM sends a message to QM that a file was cached. 
This is only to verify proper communication.


Comments:

1) We note that all of the above (except the communication with HPSS)
coincide with CORBA messages between the modules.  Thus, the easiest
tool to use is probably the Object/Observer tool that Shigeki pointed
out.

2) The Object/Observer tool has a dynamic visualization capability.
Perhaps this could be used to monitor what's going on dynamically.
I sent an email request to find out more about the visualization tool
but there was no answer so far.

3) A static display of the above for analysis will require some work
where for each query all the actions of caching and purging files need
to be collected.  I agree here with Ed's comment that putting the log
information into some database, will permit retrieval of the desired
information by issuing queries against the database.

Arie.