What basic info is needed
in troubleshooting NORAD problems?
- NORAD version
- Are NORAD server agents running
on Windows?
- Version of Windows
- Was it installed with
admin privs?
- Are NORAD server agents running
on Unix?
- What is the Unix platform
(AIX, HPUX, Solaris, Linux, Tru64, etc...)
- What is the Unix version?
- Is the OS 64-bit? Is
the database 64-bit, also?
- Database version
- Client errors - Get screenshots,
preferably pasted into a Word document.
- Unix errors - For lengthy output,
copy/paste screen output into a text file or Word document.
- A clear description of "what
is going wrong" with step-by-step
instructions on how to replicate the problem.
How do you change the default port numbers?
1) Stop the Surveillance
Server.
2) Go to cfg directory.
3) Run the following command (forward slashes for Unix):
..\bin\cfgedit dbgscom.cfg -port=[port#],CONN01
Replace [port#] with the port you want.
4) You should get a prompt asking you if you want to make
the changes. Type in 'y' and hit ENTER
5) Start Surveillance Server.
EXAMPLE:
C:\Program
Files\Bradmark\NORAD 2.0\Surveillance\cfg> ..\bin\cfgedit
dbgscom.cfg -port=15000,CONN01
File dbgscom.cfg Opened!Original values:
LISTENERS/CONN01
CONNECTSTRING=STRING,SOCKET,STREAM,SERVER,0,12203
LISTENERS/ENUM01
CONNECTSTRING=STRING,SOCKET,DGRAM,ENUM,0,12204,CONN01New Values:
SOCKET,STREAM,SERVER,0,15000
SOCKET,DGRAM,ENUM,0,15001,CONN01
Do you wish to proceed with update? (y/N)
y
Updated value:
LISTENERS/CONN01
CONNECTSTRING=STRING,SOCKET,STREAM,SERVER,0,15000
LISTENERS/ENUM01
CONNECTSTRING=STRING,SOCKET,DGRAM,ENUM,0,15001,CONN01File Closed! |
MSDE shipped with NORAD for reporting
This URL is a nice summary of the features and limits of the
MSDE database shipped with NORAD. http://www.teratrax.com/articles/msde_vs_sql_server.html
MSDE Limitations
MSDE is intended for single user or small workgroup environments.
The following are some of the MSDE limitations in comparison
with SQL Server:
No Enterprise Manager
No Query Analyzer
No Index Tuning Wizard
Only 2GB RAM
Only 2GB database size limit
Only 2 CPUs
Only five concurrent batch workloads or 25 concurrent connections
for websites
No Database Server Failover Support
No Full-text search
No SQL Server Profiler
No Import and Export Wizards
No OLAP
No English Query
No SQL Books Online
No Full or Bulk-Logged recovery model support (only simple)
Back
to top
What operators are available
when creating a rule? Surveillance Operators
Logical Operators
Operator | Purpose |
---|
AND
|
Returns TRUE if both conditions
are TRUE;
otherwise returns FALSE.
|
OR
|
Returns TRUE if either condition is TRUE;
otherwise
returns FALSE.
|
Numeric Operators
Operator | Purpose | Example |
---|
=
|
Equal to a numeric value.
|
ts# = 5
|
!=
|
Not equal to a numeric value.
|
logpct != @VALUE@
|
>
<
|
Greater than a numeric value.
Less than a numeric value.
|
CPU_RATE > @VALUE@
MergeStatus < 0 |
>=
<= |
Greater than or equal to a numeric value.
Less than or equal to a numeric value.
|
LOG_PCT_USED >= @VALUE@
time_blocked <= @VALUE@ |
IN
|
Equal to any number in set.
(Can be tricky for numbers with
decimals.)
|
IN (10,20,30 )
|
NOT IN
|
Not equal to any number in set.
(Can be tricky for numbers with
decimals.)
|
NOT IN (32,64,128)
|
INLIKE |
Equal to any number in set, with pattern matching.
|
INLIKE (1%,2%,3% )
|
NOT INLIKE |
Not equal to any number in set, with pattern matching.
|
NOT INLIKE( %0, %5 )
|
String Operators
Operator | Purpose | Attributes | Example |
---|
=
|
Equal to a string value.
|
|
status = "ONLINE"
|
!=
|
Not equal to a string value.
|
|
status != "NORMAL"
|
LIKE
|
Equal to a string value.
|
- pattern matching
- case insensitive
|
login_status like "Password%"
|
NOT LIKE
|
Not equal to a string value.
|
- pattern matching
- case insensitive
|
ACTIVE_CONN_STATE not like "%active%"
|
CONTAINS
|
Equal to a string or substring value.
|
- pattern matching
- case insensitive
|
STATUS contains "not running"
|
NOT CONTAINS
|
Does not equal to a string
or substring value.
|
- pattern matching
- case insensitive
|
NAME not contains "replication_truncation_point"
|
IN
|
Equal to any string in set.
|
|
NAME in ("SYSTEM", "RBS") |
NOT IN
|
Not equal to any string in set.
|
|
POOL not in ("java pool", "large pool") |
INLIKE
|
Equal to any string in set.
|
- pattern matching
- case insensitive
|
CLASS inlike ("%block" , "undo%") |
NOT INLIKE
|
Not equal to any string in set.
|
- pattern matching
- case insensitive
|
MetricName not inlike ("%Locks", "%sessions") |
Notes
- Pattern matching operators support
the % (percent) for a multiple
character wildcard,
and _ (underscore) for a single character wildcard.
- There is no BETWEEN.
- Parenthesis can be used for precedence
Why does the OS module require the
root.sh script to be run?
On most Unix/Linux systems, the data facilities
for OS statistics are restricted to root only. Since the OS module needs
to access
these data sources, it will also require root.
Requiring root privilege for the entire NORAD
installation would be overkill, so the root.sh script only changes
the ownership of
two programs that need the restricted facilities:
-
dbgsam, which needs access to ping
(raw-socket) facility.
-
dbgias, which needs access to OS statistics.
These programs only use root privileges when needed.
After root operations, dbsam and dbgias immediately lower
their permissions to the levels of the logged-in user
using seteuid() and setguid().
Without this root access, Surveillance should still run, but several
collection windows in the client will open up with no data.
Back
to top
What settings are available in FXCONFIG.ORA?
Setting |
Description |
Online Management |
ALLOW_TEMP_PK? |
Reorg will automatically add a PK on a non-PK table that
has a unique index on NOT NULL columns. This provides SAP
support. Default is FALSE.
Note: If an error occurs during the online reorg, the
temp PK will not be dropped. To recover, perform an online
structure change to the PK, but keep the underlying unique index.
|
DEBUG_EVENTS? |
The format of the value is "0,X1,X2,...,XN,0",
where X1,...,XN are the event numbers. This facilitates recovery testing.
There
are events 100, 200, 300,..., 2700.
The key debug events are 300, 600, 1600, and 2600.
For example, this sets debug event 600:
DEBUG_EVENTS? 0,600,0
This sets debug event 2700:
DEBUG_EVENTS? 0,2700,0
Each of these events will fake an instance
failure at a certain point in the Switchover process.
|
DESIRED_INTERVAL_COUNT? |
blah |
ENABLE_ROWID_REPLICATION? |
Allows reorg of tables without a primary key
or a unique index |
ENABLE_TBL_REORG_CALC? |
Enables the extent calculator on the Table Reorg dialog.
Setting this to TRUE will increase the time it takes for
the Reorg dialog to load the table. The default is FALSE. |
INITIAL_INTERVAL_SIZE? |
By default, the initial interval size is 1/10 the number of rows
(NUM_ROWS from the optimizer statistics on the table).
This makes the Replication Phase faster.
A larger interval size makes the interval copy more efficient,
up to a certain point, but a smaller interval size may be needed
to avoid "ORA-01555: Snapshot too old" errors with undo.
DBControl Online automatically reduces the interval size
if ORA-01555 is encountered, and it automatically increases
the interval size if it thinks it can safely do so, to improve
performance.
If INITIAL_INTERVAL_SIZE is not set, then the initial
interval will be set to...
(ESTIMATED TABLE SIZE / DESIRED_INTERVAL_COUNT)
rounded to a multiple of MINIMUM_INTERVAL_SIZE
The interval will never be
smaller than MINIMUM_INTERVAL_SIZE.
MINIMUM_INTERVAL_SIZE will be divided by 10 for tables with a LONG or LONG RAW column. |
KEEP_CONSTRAINT_NAMES? |
blah |
KEEP_INDEX_NAMES? |
Keeps index names for constraints the same after reorg. Normally,
constraint PK_T becomes PK_T__ after reorg and the index name would stay
PK_T. But if this is set to TRUE, the constraint name will become
PK_T__ or PK_T$_, and it will never be PK_T again. Default is FALSE.
Note: Normally, if this option is not set, the constraint name goes back
to the original name after a second reorg.
|
KEEP_TABLE? |
Keep the old table after reorg (_OLD). If this is set, the _OLD table must be removed
manually before the table can be reorged again.
Default is FALSE.
If the old table is in a dictionary managed tablespace and it has thousands of extents,
a DROP TABLE would be an expensive operation. Be keeping the old table,
you prevent the DROP TABLE command after the new table comes online.
You can then drop the old table manually at a later time. |
MAX_CHANGES_BEFORE_SAMPLING? |
Before we sample (for 10 seconds) to see if
the amount of DML activity is too great for Switchover,
we check whether there are too many changes to even begin
sampling. Default is 4000. |
MAX_FINAL_CHANGES? |
blah |
MAX_RUNNING_JOBS? |
blah |
MINIMUM_INTERVAL_SIZE? |
blah |
NUM_RETRY_ORA_54? 10? |
This sets the number of times to retry when an ORA-00054
(resource busy accquire with NOWAIT specified) is encountered.
|
ORACLE_COMPLETE_REFRESH? |
Forces a single refresh, instead of copying the data in pieces. |
PARALLEL_DEGREE? |
blah |
SAMPLE_INTERVAL_SECONDS? |
blah |
SEND_ALERTS? |
blah |
SORT_AREA_SIZE? |
Sets sort_area_size (Oracle parameter) for DBControl's session only.
For example, SORT_AREA_SIZE? 100000000 sets the sort_area_size to
100,000,000 bytes. |
Tracing |
SETTRACE? |
This parameter is not listed in fxconfig.ora by
default. It needs to be added. |
TRACE_INFO? |
Shows trace info on the GetOraEnvInfo request. (Added in a 2.0.9 patch). |
TRACE_ONERROR? |
blah |
TRACE_PROCS? |
Show the internal procedures that are called. |
TRACE_QUERIES? |
Shows the actual SQL queries that are sent. (A lot of them, at least). |
TRACE_TOPROGRESS? |
blah |
Why does online reorg fail on a table with only a UK, even with ALLOW_TEMP_PK
set to TRUE?
DBControl Online cannot add a temporary primary key (PK) on top of the UNIQUE index that the unique key (UK) is currently using.
That is, the UK and temporary PK cannot both use the same index, and we cannot drop the UK without losing the index on Oracle 9.0 and lower.
Back to top
|