I have a different strategy. I use the simplest, fastest SELECT
statements that can get the raw information out of the TSM database,
then move the individual tables as far away from the production TSM
server as possible to another computer. There I use my analysis and
reporting tool (SAS, SPSS, gnuS, something your brother in law wrote in
Lisp, choose your favorite) to do all those nasty joins and make
everything look nice. If my analysis does something awful such as
attempt something that will never end or divide by zero (Who, me? A bug
in a program?) it doesn't affect TSM production use.
Scripts on the analysis machine fetch the tables using the dsmadmc
command, so moving the data around is very little hassle. The idea is to
minimize the effect your monitoring and analysis activities have on
production use of the server for backing up and restoring files.
Roger Deschner University of Illinois at Chicago rogerd AT uic DOT edu
=Four-wheel-drive just means getting stuck in more inaccessible places.=
On Tue, 2 Aug 2005, Mark D. Rodriguez wrote:
>Debbie,
>
>I too was having troubles decoding exactly what it was you were trying
>to get from your select statement. My assessment was about the same as
>Andrew's, so I am not going to repeat his advice. This is a rarity but
>twice in the same day I have seen a need for the TSM Data Warehouse
>product that comes with TSM. I think the information you are trying to
>get would be better kept in the Data Warehouse where it could be
>retrieved from a real database without putting a major load on the TSM
>server itself.
>
>--
>Regards,
>Mark D. Rodriguez
>President MDR Consulting, Inc.
>
>===============================================================================
>MDR Consulting
>The very best in Technical Training and Consulting.
>IBM Advanced Business Partner
>SAIR Linux and GNU Authorized Center for Education
>IBM Certified Advanced Technical Expert, CATE
>AIX Support and Performance Tuning, RS6000 SP, TSM/ADSM and Linux
>Red Hat Certified Engineer, RHCE
>===============================================================================
>
>
>
>Debbie Bassler wrote:
>
>>Thanks, Andy,,
>>
>>I ran the script as a server script. I was not able to cancel the session,
>>we had to bounce the server.
>>
>>Is there any information on how to code scripts for TSM? If I remember
>>correctly, it's a btrieve database. Currently, I've only dealt with Oracle
>>databases and know PL/SQL code.
>>
>>Thanks again,
>>Debbie
>>
>>
>>
>>
>>
>>Andrew Raibeck <storman AT US.IBM DOT COM>
>>Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
>>08/02/2005 02:28 PM
>>Please respond to "ADSM: Dist Stor Manager"
>>
>>
>> To: ADSM-L AT VM.MARIST DOT EDU
>> cc:
>> Subject: Re: [ADSM-L] Cancel Session
>>
>>
>>WOW!
>>
>>First, exactly how did you invoke this script? For example, did you start
>>a command line admin session, then run "macro <yourscriptname>"? Or did
>>you run it via some batch file or other external script? What happens if
>>you try to issue "cancel session" again for this hung session?
>>
>>Second, I am trying to understand your SELECT command and figure out what
>>you are trying to go after. The best I can tell so far is that you are
>>really trying to roll up backup information for stuff that occurred
>>between yesterday at 18:00 through today at 08:00. In that case, your date
>>criteria could be made a little simpler:
>>
>>a.end_time >= timestamp(current_date - 1 day, '18:00:00')
>>and
>>a.end_time <= timestamp(current_date, '08:00:00')
>>
>>I'm not certain, though, that this will give you what you want. You are
>>combining records from the SUMMARY and ACTLOG tables, but even with the
>>WHERE criteria you have, I don't think you are guaranteed to get accurate
>>results. For example (and this is just one), the ACTLOG records are not
>>filtered by date, so you could get multiple records from the actlog where
>>the message ID and schedule name match your other criteria.
>>
>>Lastly, try running these commands:
>>
>>select count(*) from actlog
>>select count(*) from summary
>>
>>Multiply the results together... that will give you the total number of
>>records in the table that TSM will have to create before it does any
>>filtering... depending on how long you keep activity log and summary
>>records, that could be a very large number, resulting in a very long
>>processing time for the command.
>>
>>A compromise might be to use multiple SELECT statements to get the
>>information you need, rather than trying to roll everything into a single
>>SELECT statement. Either that, create a script/program that extracts the
>>information from each table separately, then rolls the results up into the
>>single report you want... that might be faster than having TSM do the work
>>on the joined tables.
>>
>>Regards,
>>
>>Andy
>>
>>Andy Raibeck
>>IBM Software Group
>>Tivoli Storage Manager Client Development
>>Internal Notes e-mail: Andrew Raibeck/Tucson/IBM@IBMUS
>>Internet e-mail: storman AT us.ibm DOT com
>>
>>The only dumb question is the one that goes unasked.
>>The command line is your friend.
>>"Good enough" is the enemy of excellence.
>>
>>"ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU> wrote on 2005-08-02
>>08:16:11:
>>
>>
>>
>>>This is the script which is executing;
>>>
>>>set sqldatetimeformat usa
>>>set sqldisplaymode wide
>>>select
>>>a.entity,a.successful,b.msgno,substr(b.message,posstr(b.
>>>message,'transferred:')+14,8)as
>>>Bytes_Backedup, substr(cast(min(a.start_time)as varchar(10)),1,10) as
>>>Start_Date, substr(cast(min(a.start_time)as varchar(16)),12,12) as
>>>Start_time,substr(cast(max(a.end_time)as varchar(16)),12,12) as
>>>End_Time,substr(cast(min(a.End_time)as varchar(10)),1,10) as End_Date,
>>>substr(cast(max(a.end_time)-(min(a.Start_time)) as varchar(10)),3,5) as
>>>Elapsed_Time,sum(a.mediaw) as media_Wait,sum(a.affected)as
>>>number_files,sum(a.failed) as Number_Failed,sum(a.bytes) as
>>>Bytes_Transferred, max(a.number)as Number from summary a, actlog b where
>>>b.msgno= 4961 and a.schedule_name = b.schedname and
>>>(date(a.end_time)=current date -1 days and
>>>
>>>
>>time(a.end_time)>='18.00.00')
>>
>>
>>>or( date(a.end_time)=current date) and time(a.end_time)<='08.00.00' and
>>>a.activity = 'BACKUP' group by a.entity, a.successful,b.msgno,b.message
>>>
>>>The bytes sent, bytes received and wait time are all 0.
>>>
>>>I used a predefined script, written by someone else, and wasn't aware it
>>>didn't work. I should have taken a closer look.
>>>
>>>Thanks for all the input,
>>>Debbie
>>>
>>>
>>>
>>>
>>>
>>>Andrew Raibeck <storman AT US.IBM DOT COM>
>>>Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
>>>08/02/2005 10:51 AM
>>>Please respond to "ADSM: Dist Stor Manager"
>>>
>>>
>>> To: ADSM-L AT VM.MARIST DOT EDU
>>> cc:
>>> Subject: Re: [ADSM-L] Cancel Session
>>>
>>>
>>>Look through your activity log and try to figure out the last command
>>>
>>>
>>that
>>
>>
>>>your script executed (i.e., exactly what command did the script issue
>>>
>>>
>>that
>>
>>
>>>is hung?)
>>>
>>>The "run" state would suggest that the TSM server is doing processing of
>>>
>>>
>>a
>>
>>
>>>long-running command. If you do "query session" multiple times, do you
>>>
>>>
>>see
>>
>>
>>>any of the counters for the hung session changing?
>>>
>>>You could search the IBM TSM web site
>>>
>>>http://www-306.ibm.
>>>com/software/sysmgmt/products/support/IBMTivoliStorageManager.html
>>>
>>>for "cancel session" to see if any hits match your scenario.
>>>
>>>If that doesn't yield anything useful, I'd suggest contacting IBM
>>>technical support for further assistance.
>>>
>>>Regards,
>>>
>>>Andy
>>>
>>>Andy Raibeck
>>>IBM Software Group
>>>Tivoli Storage Manager Client Development
>>>Internal Notes e-mail: Andrew Raibeck/Tucson/IBM@IBMUS
>>>Internet e-mail: storman AT us.ibm DOT com
>>>
>>>The only dumb question is the one that goes unasked.
>>>The command line is your friend.
>>>"Good enough" is the enemy of excellence.
>>>
>>>"ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU> wrote on 2005-08-02
>>>06:13:37:
>>>
>>>
>>>
>>>>I executed a script, yesterday, and it is still executing. The session
>>>>won't cancel it because it's in a run status. Is there a way to force
>>>>
>>>>
>>a
>>
>>
>>>>session to cancel?
>>>>
>>>>Thanks,
>>>>Debbie
>>>>
>>>>
>>
>>
>>
>
|
