Variable | Explanation |
SQL |
TOP_N_SQL |
How many items to list up in the "Top N"
blocks of the report for Wait objects and SQL statements |
TOP_N_WAITS |
EXC_PERF_FOR |
This can be used to exclude objects of given users from
some statistics. For example, it makes no sense to print long
execution plans for statements generated by the system and just using
system objects, since we will never go to tune these. In most cases,
the preset of "SYS SYSTEM " will be fine. If you need to
change this, beware its case sensitivity (must be all uppercase for
current Oracle releases) |
START_ID |
These two values specify the snapshot
interval to consider. Setting the value of zero for both instructs
OSPRep to automatically evaluate the interval - it will then find the
latest snapshot available and consider all snapshots since the last
instance startup for the report. Leaving the START_ID on
zero but specifying a different END_ID will take the
largest possible interval up to the given END_ID . Setting
both values to a value <> 0 will take the specified interval.
Be careful not to set START_ID to a non-zero value but
leaving END_ID on zero - this would mean just to use the
same Snapshot for both, start and end. These two values can be
overridden on the command line, using
the parameters -b (for the "Begin" snapshot) and
-e (for the "End" snapshot). |
END_ID |
SKIP_DBSTART_ID |
Skip data from the snapshot taken within the first hour
after DB startup. Useful to ignore the load from the warmup phase of
the DB. This value will be ignored when you explicitely set the
START_ID e.g. at the command line. |
MAX_REP_INTERVAL |
These two values as well specify the snapshot
interval to consider, but in a different manner. First, the
MAX_*_INTERVAL only applies if START_ID=0 .
In this case OSPRep considers the snapshot interval from the
END_ID value and subtracts the amount of days specified
by MAX_*_INTERVAL to find the starting point. Configure
this feature with the *_REP_* keyword for the reports,
and with the *_CHART_* keyword for the charts. Setting
the value of zero instructs OSPRep to evaluate the snapshot interval
as described above using START_ID and END_ID
settings. |
MAX_CHART_INTERVAL |
Features |
MK_INSTEFF |
The "Instance Efficiency Percentages" provide an overview
of some important ratios, such as "Buffer HitRatio", "Library HiRatio"
and "In-Memory Sort" and should normally be included with the report. |
MK_TOPWAITS |
Whether to display the "Top N Wait Events" block. |
MK_ALLWAITS |
Whether to display the "All Wait Events" and/or
"Background Waits" segment. In normal cases, you will not need the "All
Wait Events" and "Background Waits" segments (that's why they are
turned off by the default configuration), but they may be useful for
closer investigations. |
MK_BGWAITS |
MK_WAITOBJ |
Whether to list up the Top N Wait objects (see
TOP_N_WAITS above). This is only possible if you installed
the additional collector plugin shipped
with OSPRep - otherwise the setting of MK_WAITOBJ will
have no visible effect. |
MK_TOPSQL |
List up the Top N SQL statements at all? If set to
MK_TOPSQL=1 , four blocks will be included: Top N SQL by
Gets, by Reads, by Executions and by Parse Calls. If
MK_TOPSQL=0 , the setting for MK_EP will have
no effect at all (since execution plans are always connected to SQL
statements). |
MK_TABS |
Generate basic table statistics |
MK_EP |
If your snapshots are taken with level 6 or higher, the
execution plans are captured with each snapshot and can be displayed
together with the respective SQL statement in the TOP_N_SQL .
block. With the MK_EP parameter you can suppress these
execution plans even if they are available (by setting it to
MK_EP=0 ). However, if you only collect level 5 data with
your snapshots (or if you set MK_TOPSQL=0 , see above),
MK_EP=1 will have no effect at all. |
MK_TSIO |
The "TableSpace IO Summary Statistics" and
"File IO Summary Statistics" generally list up the same values, as long
as each TableSpace consists of exactly one datafile. In this case, you
may want to turn off one of these two stats (which normally will be the
TSIO, since FIO additionally lists up the file names).
The only other difference is: TSIO stats are ordered by IOs, while FIO
stats are ordered by TS and file names. |
MK_FIO |
MK_DBWR |
Here you define whether you want to include (1)
the DBWriter/LogWriter statistics in your report or not (0). These statistics
may help you if you assume problems with the DBWR/LGWR, since they give more
details on their activity. |
MK_LGWR |
MK_SEG_LR |
These switches are for the segment statistics
(Statspack Level 7) introduced with Oracle 9.2 and will have no effect
for Oracle < 9.2 reports. Valid settings are 0 and 1 to switch the
selected block on or off in the report: LR = Logical Reads, PR = Physical
Reads, BUSY = Buffer Busy Waits, LOCK = Row Lock Waits and ITL = ITL
Waits per segment. These reports are always ordered by waits DESC. |
MK_SEG_PR |
MK_SEG_BUSY |
MK_SEG_LOCK |
MK_SEG_ITL |
MK_INSTACT |
The "Instance Activity" segment of the report can be boring
sometimes, so you may suppress it here and just activate it if you
really need it ;-) |
MK_RECO |
The "Instance Recovery Stats" segment gives you information
on estimated recovery times, blocks etc.. |
MK_SPSTAT |
Whether to include the short common Shared Pool Stats |
MK_BUFFP |
The "Buffer Pool Statistics" segment gives
statistical information for all available pools (e.g. Default, Keep, etc.),
while the "Buffer Wait Statistics" report on waits for the different
classes (such as "data block" or "undo header"). |
MK_BUFFW |
MK_PGAA |
Whether to include the "PGA Aggregat" and
"PGA Memory" segments of the report. |
MK_PGAM |
MK_ENQ |
The "Enqueue Activity" segment of the report contains
information on for which queue types have been waits encountered |
MK_USS |
Except for special occasions, the "Undo Segment Statistics"
(MK_USSTAT ) will not be needed (especially when using
automatic undo management). The "Undo Segment Summary" table can also
be turned off, if not needed, with the MK_USS=0 statement. |
MK_USSTAT |
MK_RSSTAT |
Whether to create the "RBS Stats" and
"RBS Storage" segments of the report. These Rollback statistics are
quite useless if your database uses automatic Undo management. |
MK_RSSTOR |
MK_LACT |
"Latch Activity" will only be needed for closer investigation,
so this option is turned off by the default configuration. When you
will need this, the hints/online help of OSPRep will indicate it to you. |
MK_LMS |
The "Latch Miss Sources" report is only needed by Oracle
staff. The reason for this is quite easy: they are the only possible
people to understand it, if anybody does. So you may turn it on just for
fun (to see how it looks like), but for your normal reports you can
leave it off. Besides: if you already turned the "Latch Activity"
(MK_LACT ) off, the "Latch Miss Sources" will be
suppressed anyway. |
MK_CACHSIZ |
As introduction to the Cache Stats below (see
MK_DC and MK_LC ), this switch enables
a short Cache Size table. |
MK_DC |
Whether to create the "Dictionary Cache" and
"Library Cache" segments of the report, containing statistics and
ratios on usage of these areas. |
MK_LC |
MK_SGASUM |
This is for the "SGA Memory Summary" and
"SGA Breakdown Difference" segments. If you set MK_SGASUM=0 ,
the setting for MK_SGABREAK will be ignored and handled
as if it was set to 0 as well. |
MK_SGABREAK |
MK_RLIMS |
With the MK_RLIMS statement you can switch the
Resource Limits Report on (1) or off (0). If switched off, of course no
resource limits reports will be generated at all. If set to 1, this
report segment will be built only if you have resource limits enabled. |
MK_IORA |
Whether the "Initialization Parameters" segment should be
included with the report. This block lists up all initialization
parameters as used with the init.ora file, and the changes
made between the first and the last snapshot (e.g. with the
ALTER SYSTEM command). |
Go4Colors |
WR_* |
These are the values used for warnings (WR_* )
and alerts (AR_* ) in the Go4Colors
mode (see there for details). |
AR_* |