UCMDB and UD Practitioners Forum (Previously CMS)
cancel

UCMDB 10.10 DFP PostgreSQL database blow

Highlighted
sv_benesh
Super Contributor.

UCMDB 10.10 DFP PostgreSQL database blow

Hi All,

 

The table pg_toast_1915257  is  bloating.

 

What to do in this case? How to optimize the database?

 

dataflowprobe=# SELECT pg_size_pretty( pg_database_size( 'dataflowprobe' ) );
 pg_size_pretty
----------------
 42 GB

 

dataflowprobe=# SELECT relname AS name, relfilenode AS oid, (relpages / 128) as size_mb, reltuples  as count FROM pg_class ORDER BY relpages DESC;


                              name                              |   oid          |   size_mb |    count
----------------------------------------------------------------+---------+
 pg_toast_1915257                              | 1915260 |   41084    |       142671
 pg_toast_1915257_index                 | 1915262 |     664       |       228733

Best Regards, Sergy.
8 REPLIES
Dima Gomel
Acclaimed Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

Please check that probe has no unsent results. It culd be done from the probe JMX at http://probe:1977 -> TaskResult Destributer method.

Please also check that you have no OOM situation. In this case .hprof will be created in ../UCMDBServer/bin folder.

Regards
-Dmitry Gomel, PMP
If you find that this or any post resolves your issue, please be sure to mark it as an accepted solution.
Click the Like button at the bottom to say 'Thanks'.
sv_benesh
Super Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

I cat find the file .hprof

 

C:\hp\UCMDB\UCMDBServer\bin>dir /a
 Volume in drive C 
 Volume Serial Number 

 Directory of C:\hp\UCMDB\UCMDBServer\bin

04/17/2014  09:21 PM    <DIR>          .
04/17/2014  09:21 PM    <DIR>          ..
03/04/2014  12:31 PM             2,238 configure$3.ico
03/04/2014  12:31 PM               562 configure.bat
03/04/2014  12:30 PM            62,464 DDJDBCAuth05.dll
03/04/2014  12:30 PM            54,272 DDJDBCx64Auth05.dll
03/08/2014  05:37 AM     3,769,696,611 java_pid11144.hprof
03/10/2014  01:06 PM     3,430,624,412 java_pid11720.hprof
03/05/2014  09:45 AM       739,981,610 java_pid2176.hprof
03/04/2014  12:30 PM    <DIR>          jre
03/04/2014  12:31 PM             8,192 LWJDPAPI.dll
03/04/2014  12:31 PM             2,238 server$1.ico
03/04/2014  12:31 PM             2,414 server.bat
03/04/2014  12:31 PM             2,238 server.ico
03/04/2014  12:31 PM             2,238 status$2.ico
03/04/2014  12:31 PM                34 status.bat
03/04/2014  12:30 PM               397 wrapper-license.conf
03/04/2014  12:31 PM               361 wrapper-platform.conf
03/04/2014  12:30 PM             6,574 wrapper.conf
03/04/2014  12:31 PM           195,072 wrapper.dll
03/04/2014  12:31 PM           371,200 wrapper.exe
              18 File(s)  7,941,013,127 bytes
               3 Dir(s)  37,742,825,472 bytes free

 But I see that there is unsent data.

 

Attributes

Name	Value	Description
Active	true	Attribute exposed for management
NumberOfCurrentUnsentResults	23402	Attribute exposed for management

 

How to find out the reason why the data is not sent?

 

Best Regards, Sergy.
Dima Gomel
Acclaimed Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

I do see some:

 

You  have OOM situation. I'd recommend open support case ASAP.

Regards
-Dmitry Gomel, PMP
If you find that this or any post resolves your issue, please be sure to mark it as an accepted solution.
Click the Like button at the bottom to say 'Thanks'.
sv_benesh
Super Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

Thank You.

 

Please tell me more about it (OOM situation) or link how to decision OOM.

 

Best Regards, Sergy.
Dima Gomel
Acclaimed Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

Here you go:

http://en.wikipedia.org/wiki/Out_of_memory

Regards
-Dmitry Gomel, PMP
If you find that this or any post resolves your issue, please be sure to mark it as an accepted solution.
Click the Like button at the bottom to say 'Thanks'.
sv_benesh
Super Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

I mean how to solve in this server.

 

Total ram 8Gb

Aviable 4 Gb

 

wrapper-platform.conf

# Initial Java Heap Size (in MB (Xms))
wrapper.java.initmemory=1024

# Maximum Java Heap Size (in MB (Xmx))
wrapper.java.maxmemory=4096

Best Regards, Sergy.
Dima Gomel
Acclaimed Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

Could you please show jvm.statistics.log that includes 3/10 or 3/8 when it happend?

Regards
-Dmitry Gomel, PMP
If you find that this or any post resolves your issue, please be sure to mark it as an accepted solution.
Click the Like button at the bottom to say 'Thanks'.
sv_benesh
Super Contributor.

Re: UCMDB 10.10 DFP PostgreSQL database blow

Unfortunately files for this dates is not. This server has been cloned and log files were removed.

 

In probe-error.log

 

<2014-04-18 09:08:15,500> [ERROR] [TaskResult Grouping handler] (LocalTaskResultsDistributer.java:140) - Failed sending Task Result for job: MZ_EPM_BY_OTHER_by_agent_Network_Host Connection by Shell to server, will try on next Result Grouping handling

.............................................

<2014-04-18 09:10:39,697> [ERROR] [TaskResult Grouping handler] (ProbeTaskResultsSender.java:312) - Probe Gateway Failed storing results of task MZ_EPM_BY_OTHER_by_agent_Network_Host Connection by Shell job MZ_EPM_BY_OTHER_by_agent_Network_Host Connection by Shell from probegw dfp_server timestamp=1397732814250 objectsToAddOrUpdate Size=2858 objectsToDelete Size=45 The Exception:
org.postgresql.util.PSQLException: ERROR: duplicate key value violates unique constraint "ddm_gw_task_results_pkey"
  Detail: Key (time_stamp, probe_mgr)=(1397732814250, DFP1_SRV) already exists.

 

 

I noticed that on another DFP NumberOfCurrentUnsentResults has been sent and changed to 0.

There is also growing database.

 

dataflowprobe=# select pg_size_pretty(pg_database_size('dataflowprobe'));
 pg_size_pretty
----------------
 11 GB

 

dataflowprobe=# SELECT relname AS name, relfilenode AS oid, (relpages / 128) as size_mb, reltuples  as count FROM pg_class ORDER BY relpages DESC;
                              name                              |   oid   | size_mb |    count
----------------------------------------------------------------+---------+---------+--------------
 pg_toast_2606522                                               | 2606525 |   10699 |        19885
 pg_toast_2606522_index                                  | 2606527 |        175 |       384522

 

After I run in psql  'vacuum full;' DB size is decrease

 

dataflowprobe=# select pg_size_pretty (pg_database_size('dataflowprobe'));
 pg_size_pretty
----------------
 438 MB

 

 

Best Regards, Sergy.