Ilmar Kerm

Oracle, databases, Linux and maybe more

It feels like no-one is really talking about running Oracle database on a filesystem for years now. ASM seems to be the way to go. Oracle in the documentation even says that local file systems, although supported, are not recommended.

https://docs.oracle.com/en/database/oracle/oracle-database/19/ladbi/supported-storage-options-for-oracle-database.html#GUID-120C3995-F827-4D01-B955-CB1BD0457AF7

ASM works great and if you have RAC or a few dear pet databases, ASM for sure is a safe way to go – no argument from me here at all. I especially love the possibility to mirror the storage between different failure goups, that are defined by the administrator, so you can mirror the storage lets say between different hardware racks or availability zones or just different storage hardware. For high-end dear pet setups, ASM is no doubt the only way to go.

But I think the story changes for large cattle of small cloudy databases, where you have a large number of databases and you want to spend minimal effort managing a single one. After Oracle removed ASM from the database installation in 11g, now ASM requires Oracle Grid Infrastructure (Oracle Restart) installation – a separate Oracle software setup – that is, as any Oracle software, notoriously hard to maintain, patch – compared to standard Linux software. And the performance benefits promised by ASM, async IO and direct IO, are also available on a file system. True – the choice of filesystems and setups is harder to manage, but if you do your testing properly, the end result is easier to maintain. Do keep in mind, Oracle does not test running on filesystems any more and any problem is left to the OS support to resolve – which in case of using Oracle Linux, is also Oracle 🙂

There are many historical blogs about running Oracle on filesystems in the Internet, and the consensus seems to be that XFS is the best filesystem currently for running databases.

In addition, note Oracle Linux: Supported and Recommended File Systems (Doc ID 236826.1) also says this about XFS:

XFS is designed for high scalability and provides near native I/O performance even when the file system spans multiple storage devices. Beginning with Oracle Linux 7, XFS is the default file system and is included with Basic and Premier Support subscriptions, at no additional charge.

Based on this information I’ve been running my cattle on custom cloud small database instances on XFS, but quite soon started to have complaints about occasional database slowness and observed high log file sync and commit times. The important bit here – the applications in question commit a lot – they are micro/mini-services that get called a lot. Like many thousands of times per second. Before during platform testing I wasn’t really able to see any such behaviour, because i used tools like Swingbench and SLOB – and they are quite well behaved apps.

Then by accident found a post in Redhat support portal:

https://access.redhat.com/solutions/5906661

That rather recently XFS maintainers introduced a change, that non-aligned (to filesystem block size) direct IO write requests get serialized more thoroughly now. That change was introduced on purpose to avoid possible corruptions.

The change itself is here. It is just intorucing some extra waits on XFS filesystem level.

https://lore.kernel.org/linux-xfs/e2d4678c-eeae-e9df-dde7-84601a52d903@oracle.com/

By default XFS uses 4096 byte block size – so unless you use 4096 blocksize for Oracle redo logs – you will hit this issue. Using 512 block size redo logs on 4096 block size XFS filesystem, means Oracle will issue non-aligned writes for redo.

On Linux Oracle allows setting redo log block size to either 512 or 4096 bytes, but it has a protection mechanism – the redo log block size must match the storage physical sector size. Quoting from Doc ID 2148290.1.

Using 512N disks as 4K leads to atomicity problem (e.g. ASM/DB write operation thinks that 4K IO size is atomic which is not. It leads to consistency issues when there is a crash in the middle of a 4K write causing partial write).

If you use 512 sector size disks, then Oracle will allow you to only create redo logs with 512 byte blocksize.

# cat /sys/block/vdf/queue/physical_block_size
512
# cat /sys/block/vdf/queue/logical_block_size
512

SQL> alter database add logfile '/oralocal/1.rdo' size 1g blocksize 4096;

ORA-01378: The logical block size (4096) of file /oralocal/1.rdo is not
compatible with the disk sector size (media sector size is 512 and host sector size is 512)

If you want to use XFS for redo logs, you need to create XFS filesystem with 512 byte block size. This bring another issue – XFS has deprecated this small block size. It is still currently possible, but warning is issued.

# mkfs.xfs -b size=512 -m crc=0 -f /dev/vg01/redo
V4 filesystems are deprecated and will not be supported by future versions.
meta-data=/dev/vg01/redo         isize=256    agcount=4, agsize=31457280 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=0        finobt=0, sparse=0, rmapbt=0
         =                       reflink=0    bigtime=0 inobtcount=0
data     =                       bsize=512    blocks=125829120, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
log      =internal log           bsize=512    blocks=204800, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

To me this brings the following conclusions:

  • XFS can only be used for Oracle redo logs if you use storage with 4096 sector size (advanced format)
  • For 512 byte sector size storage, a different filesystem is needed. For example EXT4.
  • XFS is still fine to use for Oracle datafiles. This issue only concerns redo.

Testing the impact

How bad this issue is exactly. When I ran comparisons using SLOB, I did not notice the difference betwen XFS and EXT4 much. Looking at the storage patterns I saw that SLOB issues large writes and commits rarely, so storage sees quite large writes coming from the log writer.

But my apps make small changes and commit crazy often (yes, bad database application desgn – but I think quite common thinking for an average enterprise application developer). So I needed a new testing suite that can generate high number of IOPS with very small IO sizes. Introducting Horrible Little Oracle Benchmark.

https://gist.github.com/ilmarkerm/462d14da050fb73fb4eeed5384604f1b

This just commits a lot, after making a small change to the database. The results below do not represent any kind of maximum throughput limits, I just wanted to see the impact of using different filesystems under similar load.

Baseline… XFS bsize=4096

Load Profile                    Per Second   Per Transaction
~~~~~~~~~~~~~~~            ---------------   ---------------
      Redo size (bytes):       2,247,197.7             573.7
  Logical read (blocks):          19,760.3               5.0
          Block changes:          15,744.3               4.0
 Physical read (blocks):               0.3               0.0
Physical write (blocks):              45.6               0.0
       Read IO requests:               0.3               0.0
      Write IO requests:               3.6               0.0
         Executes (SQL):           3,923.7               1.0
           Transactions:           3,917.3
 
Event                                Waits Time (sec)      Wait   time
------------------------------ ----------- ---------- --------- ------
log file sync                    2,367,110     3340.7    1.41ms   92.8
DB CPU                                          281.9              7.8
buffer busy waits                  129,582        4.6   35.49us     .1
log file switch (private stran          10         .1    9.29ms     .0
db file sequential read                122         .1  513.16us     .0
control file sequential read            98          0  327.92us     .0

                                                Total
                                       %Time     Wait              Waits   % bg
Event                            Waits -outs Time (s)  Avg wait     /txn   time
-------------------------- ----------- ----- -------- --------- -------- ------
log file parallel write      1,182,647     0    1,166     .99ms      0.5   64.6
LGWR any worker group        1,179,342     0      592  501.58us      0.5   32.8

EXT4

Load Profile                    Per Second   Per Transaction
~~~~~~~~~~~~~~~            ---------------   ---------------
      Redo size (bytes):       4,151,622.2             571.5
         Executes (SQL):           7,267.6               1.0
           Transactions:           7,264.7
 
Event                                Waits Time (sec)      Wait   time
------------------------------ ----------- ---------- --------- ------
log file sync                    4,386,595     3114.6  710.03us   86.5
DB CPU                                          517.8             14.4
buffer busy waits                  302,823       11.1   36.71us     .3

                                                Total
                                       %Time     Wait              Waits   % bg
Event                            Waits -outs Time (s)  Avg wait     /txn   time
-------------------------- ----------- ----- -------- --------- -------- ------
log file parallel write      1,954,831     0      841  430.21us      0.4   59.8
LGWR any worker group        1,226,037     0      271  221.22us      0.3   19.3
LGWR all worker groups         330,590     0      166  502.00us      0.1   11.8
LGWR worker group ordering     318,164     0       47  146.44us      0.1    3.3

XFS with blocksize 512

Load Profile                    Per Second   Per Transaction
~~~~~~~~~~~~~~~            ---------------   ---------------
      Redo size (bytes):       4,118,949.8             571.4
         Executes (SQL):           7,213.7               1.0
           Transactions:           7,208.6
 
Event                                Waits Time (sec)      Wait   time
------------------------------ ----------- ---------- --------- ------
log file sync                    4,354,303     3127.9  718.34us   86.9
DB CPU                                          503.7             14.0
buffer busy waits                  349,311       12.5   35.76us     .3

                                                Total
                                       %Time     Wait              Waits   % bg
Event                            Waits -outs Time (s)  Avg wait     /txn   time
-------------------------- ----------- ----- -------- --------- -------- ------
log file parallel write      1,950,597     0      847  434.01us      0.4   59.7
LGWR any worker group        1,245,012     0      277  222.65us      0.3   19.5
LGWR all worker groups         330,132     0      169  512.99us      0.1   11.9
LGWR worker group ordering     318,918     0       44  139.26us      0.1    3.1

EXT4 with LVM

On previous tests I created file system on top of the raw block device, here I want to compare what happens if there is LVM in the middle.

Load Profile                    Per Second   Per Transaction
~~~~~~~~~~~~~~~            ---------------   ---------------
      Redo size (bytes):       3,412,249.5             571.3
         Executes (SQL):           5,981.8               1.0
           Transactions:           5,972.8
 
Event                                Waits Time (sec)      Wait   time
------------------------------ ----------- ---------- --------- ------
log file sync                    3,613,132     3203.5  886.64us   88.9
DB CPU                                          423.9             11.8
buffer busy waits                  301,471         11   36.60us     .3

                                                Total
                                       %Time     Wait              Waits   % bg
Event                            Waits -outs Time (s)  Avg wait     /txn   time
-------------------------- ----------- ----- -------- --------- -------- ------
log file parallel write      1,610,090     0      843  523.49us      0.4   59.0
LGWR any worker group        1,024,262     0      280  273.01us      0.3   19.6
LGWR all worker groups         270,089     0      171  631.60us      0.1   11.9
LGWR worker group ordering     310,863     0       66  210.95us      0.1    4.6

The important bits

  • Oracle RDBMS 19.16.2
  • Oracle Linux 8.7 with UEKr6 kernel 5.4.17-2136.316.7.el8uek.x86_64
  • Dedicated filesystem for redo logs
  • filesystemio_options=setall
  • No partitioning on block devices

During all tests datafiles remained at the same filesystem and the load on them was absolutely minimal. Between each test I reformatted the redo log filesystem and copied the redo log files over to the new filesystem.

I also tested both XFS and EXT4 when filesystemio_options set to asynch or none, but both of them performed similarly to XFS bsize=4096 with filesystemio_options=setall.

Conclusion

From my tests I conclude, that XFS performance degradation (when not using 4096 byte sector size storage) for high-commit rate applications is significant. The slowdown is observed as longer latency in log file sync and log file parallel write wait events.

As a replacement, EXT4 and XFS blocksize=512 perform similarly. Adding LVM in the mix, reduces the commit throughput a little. Since XFS blocksize=512 is deprecated, EXT4 is the go-to major filesystem on Linux for Oracle database redo logs.

Before going for a file system for Oracle database, do read through Oracle Linux: Supported and Recommended File Systems (Doc ID 236826.1) to understand the implications.

Here is a list of things that I think are important to monitor if you have Data Guard FSFO setup. All the mentioned things are intended for automated monitoring – things to raise alerts on.

My goal is to get as much information as possible from one place – the primary database. Since the primary database is alwas open and you can always connect your monitoring agent there.

This is about Oracle 19c (19.16.2 at the time of writing).

Here I’m also assuming Data Guard protection mode is set to MaxAvailability (or MaxProtection).

The overall status of Data Guard broker

I think the best way to get a quick overview of Data Guard health is to query V$DG_BROKER_CONFIG. It can be done from primary and it gives you a quick status for each destination.

SYS @ failtesti2:>SELECT database, dataguard_role, enabled, status
  FROM V$DG_BROKER_CONFIG;

DATABASE     DATAGUARD_ROLE     ENABL     STATUS
------------ ------------------ ----- ----------
failtesti1   PHYSICAL STANDBY   TRUE           0
failtesti2   PRIMARY            TRUE           0
failtesti3   PHYSICAL STANDBY   TRUE       16809

STATUS column is the most interesting. It gives you the current ORA error code for this destination. If it is 0 – ORA-0 means “normal, successful completion”.

Is FSFO target SYNCHRONIZED?

In case Fast-Start Failover is enabled – you need to be alerted if the FSFO target becomes UNSYNCHRONIZED. You can query it from V$DATABASE.

SYS @ failtesti2:>SELECT FS_FAILOVER_MODE, FS_FAILOVER_STATUS, FS_FAILOVER_CURRENT_TARGET, FS_FAILOVER_THRESHOLD
  FROM V$DATABASE;

FS_FAILOVER_MODE    FS_FAILOVER_STATUS     FS_FAILOVER_CURRENT_TARGET     FS_FAILOVER_THRESHOLD
------------------- ---------------------- ------------------------------ ---------------------
ZERO DATA LOSS      SYNCHRONIZED           failtesti1                                        25

-- After bouncing failtesti1

SYS @ failtesti2:>SELECT FS_FAILOVER_MODE, FS_FAILOVER_STATUS, FS_FAILOVER_CURRENT_TARGET, FS_FAILOVER_THRESHOLD
  FROM V$DATABASE;

FS_FAILOVER_MODE    FS_FAILOVER_STATUS     FS_FAILOVER_CURRENT_TARGET     FS_FAILOVER_THRESHOLD
------------------- ---------------------- ------------------------------ ---------------------
ZERO DATA LOSS      UNSYNCHRONIZED         failtesti1                                        25

Alert if FS_FAILOVER_STATUS has been UNSYNCHRONIZED for too long.

Is Observer connected?

In FSFO you also need to alert if observer is no longer present. If there is no observer, FSFO will not happen and you also loose quorum. If both standby database and observer are down – primary databases loses quorum and also shuts down.

Again, you can query V$DATABASE from the primary database.

SYS @ failtesti2:>SELECT FS_FAILOVER_MODE, FS_FAILOVER_CURRENT_TARGET, FS_FAILOVER_OBSERVER_PRESENT, FS_FAILOVER_OBSERVER_HOST
  FROM V$DATABASE;

FS_FAILOVER_MODE    FS_FAILOVER_CURRENT_TARGET     FS_FAIL FS_FAILOVER_OBSERVER_HOST
------------------- ------------------------------ ------- ------------------------------
ZERO DATA LOSS      failtesti1                     YES     failtest-observer

-- After stopping the observer
SYS @ failtesti2:>SELECT FS_FAILOVER_MODE, FS_FAILOVER_CURRENT_TARGET, FS_FAILOVER_OBSERVER_PRESENT
  FROM V$DATABASE;

FS_FAILOVER_MODE    FS_FAILOVER_CURRENT_TARGET     FS_FAIL
------------------- ------------------------------ -------
ZERO DATA LOSS      failtesti1                     NO

Alert if FS_FAILOVER_OBSERVER_PRESENT has been NO for too long.

Is the protection mode as intended?

Good to monitor to avoid DBA human errors. Maybe a DBA lowered the protection mode and forgot to reset it back to the original value. Again this information is available on V$DATABASE. Probably not needed, but it is an easy and check check.

SYS @ failtesti2:>SELECT PROTECTION_MODE, PROTECTION_LEVEL
    FROM V$DATABASE;

PROTECTION_MODE      PROTECTION_LEVEL
-------------------- --------------------
MAXIMUM AVAILABILITY MAXIMUM AVAILABILITY

Monitoring Apply and Transport lag

Probably this is one of the most asked about things to monitor about Data Guard. And the obvious way to do it is via Data Guard Broker.

First – you have ApplyLagThreshold and TransportLagThreshold properties for each database and if that is breached, Data Guard Broker will raise an alarm – the database status will change.

-- On my standby database I have ApplyLagThreshold and TransportLagThreshold set
DGMGRL> show database failtesti3 ApplyLagThreshold;
  ApplyLagThreshold = '30'
DGMGRL> show database failtesti3 TransportLagThreshold;
  TransportLagThreshold = '30'

-- I do breach both of the thresholds
DGMGRL> show database failtesti3;

Database - failtesti3

  Role:               PHYSICAL STANDBY
  Intended State:     APPLY-ON
  Transport Lag:      1 minute 38 seconds (computed 49 seconds ago)
  Apply Lag:          1 minute 38 seconds (computed 49 seconds ago)
  Average Apply Rate: 6.00 KByte/s
  Real Time Query:    ON
  Instance(s):
    failtesti3

  Database Warning(s):
    ORA-16853: apply lag has exceeded specified threshold
    ORA-16855: transport lag has exceeded specified threshold
    ORA-16857: member disconnected from redo source for longer than specified threshold

Database Status:
WARNING

-- If I query V$DG_BROKER_CONFIG from PRIMARY the status reflects that
SYS @ failtesti2:>SELECT database, dataguard_role, status
    FROM V$DG_BROKER_CONFIG;

DATABASE        DATAGUARD_ROLE         STATUS
--------------- ------------------ ----------
failtesti1      PHYSICAL STANDBY            0
failtesti2      PRIMARY                     0
failtesti3      PHYSICAL STANDBY        16809

$ oerr ora 16809
16809, 00000, "multiple warnings detected for the member"
// *Cause:  The broker detected multiple warnings for the member.
// *Action: To get a detailed status report, check the status of the member
//          specified using either Enterprise Manager or the DGMGRL CLI SHOW
//          command.

-- If I fix the transport lag
SYS @ failtesti2:>SELECT database, dataguard_role, status
    FROM V$DG_BROKER_CONFIG;

DATABASE        DATAGUARD_ROLE         STATUS
--------------- ------------------ ----------
failtesti1      PHYSICAL STANDBY            0
failtesti2      PRIMARY                     0
failtesti3      PHYSICAL STANDBY        16853

$ oerr ora 16853
16853,0000, "apply lag has exceeded specified threshold"
// *Cause:  The current apply lag exceeded the value specified by the
//          ApplyLagThreshold configurable property. It may be caused either by
//          a large transport lag or poor performance of apply services on the 
//          standby database.
// *Action: Check for gaps on the standby database. If no gap is present, tune
//          the apply services.

Really good for monitoring and the monitoring agent only needs to connect to the primary database. If status != 0, we have a problem.

OK good, but that will only tell you that the threshold was breached – but how big is the lag actually? Managers like charts. If you can query from standby database, V$DATAGUARD_STATS is your friend.

-- Has to be executed from standby database
WITH
    FUNCTION interval_to_seconds(p_int interval day to second) RETURN number
        DETERMINISTIC
    IS
    BEGIN
        -- Converts interval to seconds
        RETURN 
            extract(day from p_int)*86400+
            extract(hour from p_int)*3600+
            extract(minute from p_int)*60+
            extract(second from p_int);
    END;
SELECT name, interval_to_seconds(to_dsinterval(value)) lag_s
FROM v$dataguard_stats
WHERE name IN ('transport lag','apply lag')

Another view to get more details about the apply process is V$RECOVERY_PROGRESS. Again have to query it from the standby database itself. You can see apply rates for example from there.

What if you can not (do not want to) connect to (each) standby database, maybe it is in MOUNTED mode (I object to connecting remotely as SYS)? How to get standby lag querying from primary database?

Documented option is to query V$ARCHIVE_DEST, not great and does not work for cascaded destinations.

SYS @ failtesti2:>SELECT dest_id, dest_name, applied_scn, scn_to_timestamp(applied_scn)
  FROM v$archive_dest
  WHERE status = 'VALID' and applied_scn > 0;

   DEST_ID DEST_NAME            APPLIED_SCN SCN_TO_TIMESTAMP(APPLIED_SCN)
---------- -------------------- ----------- -----------------------------
         2 LOG_ARCHIVE_DEST_2     409036447 2023-03-11 07:40:00
         3 LOG_ARCHIVE_DEST_3     409173280 2023-03-11 07:41:27

Not really great for monitoring.

Is there a possibility to query Broker data from primary database? Currently not in a documented way.

-- Query from Broker. NB! DBMS_DRS is currently undocumented.
-- X$DRC is X$, so it will never be documented, and you can only query it as SYS/SYSDG.

SYS @ failtesti2:>SELECT value database_name, dbms_drs.get_property_obj(object_id, 'ApplyLag') apply_lag
  FROM x$drc
  WHERE attribute='DATABASE' and value != 'failtesti2';

DATABASE_NAM APPLY_LAG
------------ ------------
failtesti1   0 0
failtesti3   1174 40

-- As a comparison
SYS @ failtesti2:>SELECT dest_id, dest_name, applied_scn, cast(systimestamp as timestamp) - scn_to_timestamp(applied_scn) lag
  FROM v$archive_dest
  WHERE status = 'VALID' and applied_scn > 0;

   DEST_ID DEST_NAME            APPLIED_SCN LAG
---------- -------------------- ----------- ------------------------------
         2 LOG_ARCHIVE_DEST_2     409036447 +000000000 00:20:13.863936000
         3 LOG_ARCHIVE_DEST_3     411005382 +000000000 00:00:40.863936000

-- Comparing these numbers it seems to me that "1174 40" should be interpreted as "lag_in_seconds measurement_age_in_seconds"
-- So failtesti3 had 1174 seconds of lag measured 40 seconds ago

-- The same also works for TransportLag
SYS @ failtesti2:>SELECT value database_name, dbms_drs.get_property_obj(object_id, 'TransportLag') transport_lag
  FROM x$drc
  WHERE attribute='DATABASE' and value != 'failtesti2';

DATABASE_NAM TRANSPORT_LAG
------------ -------------
failtesti1   0 0
failtesti3   107 56

Also – my favourite actually – if you have Active Data Guard licence and the standby is open – just have a pinger job in primary database updating a row every minute (or 30s??) to the current UTC timestamp sys_extract_utc(systimestamp). And have your monitoring agent check on the standby side how old that timestamp is. It is my favourite, because it is a true end-to-end check, it does not depend on any dark unseen Data Guard inner workings and wonderings if the reported numbers by Broker are correct.

Is MRP running (the apply process)?

Monitoring lag via Broker is great, but if by DBA accident you set APPLY-OFF for a maintenance – and forget to start it again. Since administrator has told Broker that APPLY should be OFF, Broker will not raise ApplyLag warnings anymore. So if Failover happens – it will be very slow, since it also needs to apply all the missing logs.

-- To demonstrate the problem

DGMGRL> show database failtesti3 ApplyLagThreshold;
  ApplyLagThreshold = '30'

-- ApplyLagThreshold is 30s and current Apply Lag is 25 min, but status is SUCCESS!
-- This is because Intended State: APPLY-OFF
DGMGRL> show database failtesti3;

Database - failtesti3

  Role:               PHYSICAL STANDBY
  Intended State:     APPLY-OFF
  Transport Lag:      0 seconds (computed 1 second ago)
  Apply Lag:          25 minutes 3 seconds (computed 1 second ago)
  Average Apply Rate: (unknown)
  Real Time Query:    OFF
  Instance(s):
    failtesti3

Database Status:
SUCCESS

-- Same from primary, failtesti3 has status=0
SYS @ failtesti2:> SELECT database, dataguard_role, enabled, status
    FROM V$DG_BROKER_CONFIG;

DATABASE     DATAGUARD_ROLE     ENABL     STATUS
------------ ------------------ ----- ----------
failtesti1   PHYSICAL STANDBY   TRUE           0
failtesti2   PRIMARY            TRUE           0
failtesti3   PHYSICAL STANDBY   TRUE           0

-- Sure will still show Apply Lag from Broker
SYS @ failtesti2:> SELECT value database_name, dbms_drs.get_property_obj(object_id, 'ApplyLag') apply_lag
    FROM x$drc
    WHERE attribute='DATABASE' and value = 'failtesti3';

DATABASE_NAM APPLY_LAG
------------ ------------
failtesti3   1792 0

How to check that all targets are applying? One option is to actually check on standby side, if MRP0 process is running.

SYS @ failtesti3:>SELECT process
    FROM V$MANAGED_STANDBY
    WHERE process = 'MRP0';

PROCESS
---------
MRP0

-- I shut down apply
DGMGRL> edit database failtesti3 set state='apply-off';
Succeeded.

-- No MRP0 anymore
SYS @ failtesti3:>SELECT process FROM V$MANAGED_STANDBY WHERE process = 'MRP0';

no rows selected

But for automated monitoring I don’t really like to connect to each standby, especially if no Active Data Guard is in use. How to get it from primary? It gets a little tricky.

-- To check the intended state, could also go for the unsupported route.
-- APPLY-READY means APPLY-OFF
SYS @ failtesti2:>SELECT value database_name, dbms_drs.get_property_obj(object_id, 'intended_state') intended_state
    FROM x$drc
    WHERE attribute='DATABASE' and value = 'failtesti3';

DATABASE_NAM INTENDED_STATE
------------ --------------------
failtesti3   PHYSICAL-APPLY-READY

-- After turning APPLY-ON
DGMGRL> edit database failtesti3 set state='apply-on';
Succeeded.
SYS @ failtesti2:>SELECT value database_name, dbms_drs.get_property_obj(object_id, 'intended_state') intended_state
    FROM x$drc
    WHERE attribute='DATABASE' and value = 'failtesti3';

DATABASE_NAM INTENDED_STATE
------------ --------------------
failtesti3   PHYSICAL-APPLY-ON

Here could check that Broker has the intended state correct for all standby databases.

Broker also has export configuration option, that could possibly be used for some automated checks. It is a documented DGMGRL command, but the resulting XML file is placed under database trace directory and the contents of that XML file are not documented.

[oracle@failtest-2 ~]$ dgmgrl / "export configuration to auto-dg-check.xml"
DGMGRL for Linux: Release 19.0.0.0.0 - Production on Sat Mar 11 09:04:17 2023
Version 19.16.2.0.0

Copyright (c) 1982, 2019, Oracle and/or its affiliates.  All rights reserved.

Welcome to DGMGRL, type "help" for information.
Connected to "FAILTESTI2"
Connected as SYSDG.
Succeeded.

-- The result is an XML file with all interesting Broker configurations that could be used for automated monitoring

[oracle@failtest-2 ~]$ cat /u01/app/oracle/diag/rdbms/failtesti2/failtesti2/trace/auto-dg-check.xml

...
  <Member MemberID="1" CurrentPath="True" Enabled="True" MultiInstanced="True" Name="failtesti1">
    <DefaultState>STANDBY</DefaultState>
    <IntendedState>STANDBY</IntendedState>
    <Status>
      <Severity>Success</Severity>
      <Error>0</Error>
      <Timestamp>1678525406</Timestamp>
    </Status>
    <Role>
      <Condition>STANDBY</Condition>
      <DefaultState>PHYSICAL-APPLY-ON</DefaultState>
      <IntendedState>PHYSICAL-APPLY-ON</IntendedState>
    </Role>
...
  </Member>
...
  <Member MemberID="3" CurrentPath="True" Enabled="True" MultiInstanced="True" Name="failtesti3">
    <Status>
      <Severity>Success</Severity>
      <Error>0</Error>
      <Timestamp>1678525396</Timestamp>
    </Status>
    <Role>
      <Condition>STANDBY</Condition>
      <DefaultState>PHYSICAL-APPLY-ON</DefaultState>
      <IntendedState>PHYSICAL-APPLY-READY</IntendedState>
    </Role>
...
  </Member>
...

Recovery Area

Need to cover also the basics – like Recovery Area usage. When using Data Guard you have most likely set ArchivelogDeletionPolicy to Applied On (All) Standby… or shipped, so if any of the standby databases fall behind, the recovery are on primary (or other standby databases) will also start to grow. Keep an eye on that.

Useful views: V$RECOVERY_FILE_DEST (or V$RECOVERY_AREA_USAGE to see the breakdown into individual consumers).

SELECT sys_extract_utc(systimestamp) time
    , name recovery_dest_location
    , round(space_limit/1024/1024/1024) size_gb
    , round((space_used-space_reclaimable)/1024/1024/1024) used_gib
    , round((space_used-space_reclaimable)*100/space_limit) used_pct
FROM V$RECOVERY_FILE_DEST

Connecting to the alert.log topic below… if you have been slow to react on the proactive recovery area check above, always good to keep mining alert.log for ORA-19809. If that is seen, raise the highest immediate alert – recovery area is full and archiving is stopped.

Alert.log

Always a good idea to keep an eye on important messages from alert.log even if they are not directly related to Data Guard. I’ve blogged about how I mine and monitor alert.log here.

What to alert on? Anything where level < 16 also keep an eye on if comp_id=’VOS’.

System wait events

In case of using FSFO, your standby database is most likely in SYNC or FASTSYNC mode, this is having an impact on end user commit times. So good to keep an eye on commit timings.

V$SYSTEM_EVENT is a good view for this, but you have to sample it regularly and report on the deltas.

-- Need to sample this regularly. Report on deltas.
SELECT sys_extract_utc(systimestamp) sample_time, event
    , time_waited_micro_fg
    , time_waited_micro
FROM v$system_event
WHERE event in ('log file sync','log file parallel write','Redo Transport MISC','SYNC Remote Write');

<RANT/>

Please only use UTC times (or TIMESTAMP WITH TIME ZONE) when developing your internal monitoring (or any other software). This is the global time standard for our planet. It is always growing, no strange jumps, everyone knows what it is. Local times belong only to the user interface layer.

Did I miss anything?

Most likely – but please tell me! I’m very interested in getting Data Guard FSFO monitoring rock solid.

Oracle REST Data Services

Recently ORDS also added a REST interface to query Data Guard properties. I have not yet checked it out, but potential for monitoring is high – especially to get around using the undocumented features.

https://docs.oracle.com/en/database/oracle/oracle-database/21/dbrst/op-database-dataguard-databases-database-properties-get.html

At minimum need to trace what queries it is executing in the background to replicate them for custom monitoring 🙂

It is possible to use DBMS_SQL to execute dynamic SQL in multiple pluggable databases, from CDB. Oracle does provide CONTAINERS clause, but the use of it is quite limited.

In this example I execute from CDB and I’m stopping all services in all PDB-s, transactionally.

Categories