DWS Pool Urgent Purge Troubleshooting

Asset ID:	1-72-2047078.1
Update Date:	2017-11-27
Keywords:

Solution Type Problem Resolution Sure

Solution 2047078.1 : DWS Pool Urgent Purge Troubleshooting

Applies to:

Oracle Communications Performance Intelligence Center (PIC) Software - Version 4.1 and later
Information in this document applies to any platform.
This article is only about DWS pool urgent purge, it is not covering the PDU pool urgent purge.

Symptoms

DATA_CDR or DATA_IND tablespace are used at 90% or more:

[cfguser@ixp00xx-1a ~]$ ViewTbspceUsage.sh -a ixp0001_DWS

Tablespace                         Used (Mb)      Max (Mb)     Used %
------------------------------ ------------- ------------- ----------
DATA_CDR                     32935       180224          18
DATA_CONF                                  5 2048           0
DATA_IND                               13918    16384    90
DATA_LOG                                  21 2048           1
SYSAUX                                  1138    12288 9
SYSTEM                                   365    12288 3
UNDO                                    5990    16384    37

The table space usage can also be checked in the Capacity management session by filtering on the Store DFP in the Direction Outgoing, the resource Load indicator give the highest value from the CDR or the Index.

Following symptoms can also be seen:

xDR or KPI sessions have shorter or even no history
slower query response time

Changes

Is it due to all servers from the pool or only one?

Is it due to the DATA_CDR or DATA_IND tablespace?

Are the table space using all the available storage?

What time are you checking the usage?

Cause

Possible causes:

Traffic has increased since the installation
Correlation rate have decreased
The nightly jobs in charge of the purge or the xDRs compression are not working
The servers were not installed or re-installed properly

Solution

Follow this checklist:

Erase sessions which are no more used (reconstitution but also KPI sessions):
- KPI sessions are not erased when the KPI configuration are erased
- Use ViewSessionFlow.sh script to identify inactive sessions
- Identify the session using the more space with ViewObjectSpace.sh script
Check the IXP nightly job logs and especially the ones related to the compression
- Consider the data are purged only once a day so if you are checking the usage just after the maintenance window it can happen the pool is in urgent purge only in the evening.
Check if some indexes has been added or removed:
- This may impact the queries performance.
Check the tablespace capacity:
- It can happen they were not extended to the full capacity
Reduce the session duration:
- This is only the ultimate solution in case the traffic really increased, or in emergency until the previous options investigate.

It is possible to "protect" some sessions from being erased by an Urgent purge mechanism.This might be especially important for statistical sessions as those are often of higher importance for customers. It must not be applied on big xDR sessions. For this session, only normal purge will be applied.

To mark session as being excluded from Urgent Purge, execute as root:

cd_oracle_utils
./ManageUrgentPurge.sh ixp/password@IPadress/ixp Session_Name -e

Expected output:

Urgent purge is currently DISABLED for session Session_Name

References

<NOTE:1935728.1> - Troubleshooting Guides for PIC 10.1.0

Attachments

This solution has no attachment