This post is more than 5 years old

162027

November 18th, 2010 03:00

Foglight v5.5.5 tablespace monitoring when using autoextend

Basically, we've turned off monitoring using "DBO - Tablespace Used Pct" because of the autoextend implementation across all our databases.

The existing Rule appears to be based on percent used of "Allocated" space. All tablespaces will therefore alert as each crosses their thresholds prior to going 100% full and autoextending.

It appears that the only way around this is to write a new rule to evaluate relative free space in the data files. This would be done at the database level using SQL against the  SYS.dba_data_files (MAXBYTES, USER_BYTES) and SYS.dba_free_space (BYTES) definitions and doing some maths (not forgetting some degree of filtering).

Our preference would be to use the current Foglight DBO data to drive a valid rule.

Any comments  would be appreciated.

Cheers

Col

3 Posts

November 25th, 2010 09:00

Hi,

there are some failures in my specific database. First: I  have regrant the permissions into the database to my monitoring user.  Second: enabled OS monitoring. The alarm was cleared by rule.  Conclusion: The tablespace monitoring is working as expected - that's  also true for autoextensible files/tablespaces. Regards Jochen

3 Posts

November 19th, 2010 11:00

Hi Colin, I mean that the autoallocating feature is observed in the oracle monitoring car. That's a conclusion between oracle view and filesystem view and you should get a alarm if the underlaying filesystem is near by xxx % capacity usage. I'll check this on monday in my dev environment.

1 Message

November 24th, 2010 10:00

Jochen,

Is the table space rule considering autoextend correctly then ?  I am trying to query for the space remaining in the file system for a tablespace considering autoextend.

In DBO_tablespace there is #DBO_free_pct# does this consider autoextend and the file system ?

There is also DBO_Total_Used_Space_Pct in DBO_Total_Database_Storage but not certain due to lack of documentation.

Many thanks for any comments.

November 24th, 2010 18:00

In Foglight 4 monitoring, we have had to turn off auto extend feature for tablespaces to get correct Used Pct. I was told by Quest Foglight Support that Foglight 5.5.5 may be handling it correctly as this has been a known issue with Foglight 4.

Reading Colin's post, I am now worried as I have been trying to transition to Foglight 5.5.5 monitoring to resolve the auto-extend issue with Foglight 4. Jochen, could you please update your post as soon as possible with whatever you have found in your Dev Environment?

4 Posts

November 24th, 2010 21:00

Colin,

The Table Space rule uses a Derived metric in its condition. the derived metric uses the following code to construct the pct value

         try {
                result=0;
                if ((#auto_os_datafile_count# == 0) && (#allocated_space_used_pct#!=null))
                    result = #allocated_space_used_pct#;
                if ((#DBO_free_pct#!=null)&&(#auto_os_datafile_count# > 0))
                    result = 100-#DBO_free_pct#;

                return result; // so that the alarm will not be activated
            }
            catch (Exception e) {
                Dbwc_Common_Oracle_ExceptionInvestigation ("DBO_allocated_space_used_pct_for_rules", e,
                    [],
                    [
                        "#auto_os_datafile_count#" : #auto_os_datafile_count#,
                        "#allocated_space_used_pct#" : #allocated_space_used_pct#,
                        "#DBO_free_pct#" : #DBO_free_pct#
                    ]
                );
            }

This should be taking into consider Oracles AutoExtend functionality

If this rule is not functioning as it should then it would be best to inform support@quest.com that we have an issue in the rule/derived metric logic

Hope this helps

3 Posts

November 25th, 2010 07:00

Hi Colin and Lee,

I got the same error for databases with autoextending tabelspaces, but not for all. I have disabled OS monitoring for some hosts where more than one database resides -> extensive memory/cpu consuming. Can you check on your side the collected values (observation) ( Alarm Message -> Instance -> Property viewer ->

e.g. num_of_files, used_growth, auto_os_datafile_count? For my database are all values "n/a". I'll test the behavior with enabled OS monitoring again and update the thread. Regards Jochen

December 8th, 2010 02:00

Hi,

sorry about the delay. Went off and installed Cartridges for 555 SP2 just to bring things up to date (Also fixes Alert Log alarm bug #ODB-2219)

The clue was in the OS Monitoring that you mentioned, and confirmed in 555 SP2 Release Notes issue #ODB-2014.

Tested the autoextend through to max size, leaving gb's still free on the disk. All good. Warning and Critical alarms fired in the appropriate places.

A much better result.

555 User Guide is a must read also.

Thanks for your time and feedback. Much appreciated.

Cheers

Col

No Events found!

Top