Sun Microsystems, Inc.  Oracle System Handbook - ISO 7.0 May 2018 Internal/Partner Edition
   Home | Current Systems | Former STK Products | EOL Systems | Components | General Info | Search | Feedback

Asset ID: 1-77-2308117.1
Update Date:2017-10-20
Keywords:

Solution Type  Sun Alert Sure

Solution  2308117.1 :   (DB40) Database instance startup slow or ASM rebalance slow on 8-socket database servers with 12.1.0.2.170718 or 12.1.0.2.170814  


Related Items
  • Exadata X6-8 Hardware
  •  
  • Exadata X4-8 Hardware
  •  
  • Exadata X5-8 Hardware
  •  
  • Exadata X7-8 Hardware
  •  
  • Oracle Exadata Storage Server Software
  •  
Related Categories
  • PLA-Support>Eng Systems>Exadata/ODA/SSC>Oracle Exadata>DB: Exadata_EST
  •  




In this Document
Description
Occurrence
 Pre-requisite Conditions
Symptoms
Workaround
Patches
History
References


Applies to:

Exadata X7-8 Hardware
Oracle Exadata Storage Server Software - Version 12.1.2.3.0 and later
Exadata X6-8 Hardware
Exadata X4-8 Hardware
Exadata X5-8 Hardware
Information in this document applies to any platform.

Description

One or both of the following issues may occur:

  1. Bug 26762083 - Database instance startup open is slower than normal, possibly causing clusterwide performance impact to sessions connected to other instances.
  2. Bug 26626720 - ASM rebalance may be very slow or appear to hang.

Occurrence

Pre-requisite Conditions

The following conditions must exist for these issues to occur:

  1. Exadata database server is an X4-8 or later generation 8-socket system.
  2. Exadata version is 12.1.2.3.0 or higher on database servers.
  3. The database home or grid infrastructure home on one or more database servers has the fix for bug 25256170 installed.
    1. This fix was included in 12.1.0.2.170718 and 12.1.0.2.170814.
    2. This fix may have been applied as an interim patch on top of earlier 12.1.0.2 releases.
    3. In a rolling software update scenario, these issues can occur after applying the fix for bug 25256170 (or a bundle patch containing it) to just the first node.

 

1. To determine if the database server is an 8-socket or 2-socket server run the following command as root on the first database server:

# dmidecode -s system-product-name
ORACLE SERVER X5-8

Only X4-8 and later generation 8-socket database servers are affected by this issue.

2. To determine Exadata version run the following command as root on database servers:

# imageinfo -ver
12.1.2.3.5.170418

 

3. To determine if the fix for bug 25256170 is installed run the following OPatch command. Run this command for both the database home and grid infrastructure home. The following example output shows the fix for bug 25256170 is installed:

$ $ORACLE_HOME/OPatch/opatch lsinventory -bugs_fixed | grep 25256170
25256170 25869760 Tue Jul 18 22:59:12 PDT 2017 TRACKING BUG FOR BDSQL 3.1.0 RELEASE ON TOP OF

 

This issue occurs on database servers with multiple loopback IP addresses (i.e. loopback addresses in addition to 127.0.0.1, like 127.0.0.2/3/4, defined and used by the database or ASM).  Multiple loopback IP addresses are created only on X4-8 and later generation 8-socket database servers running Exadata 12.1.2.3.0 or higher.

 

Symptoms

Symptoms include the following: 

  1. Bug 26762083 - Database instance startup open is slower than normal and new and existing connections to instances for the same database on other nodes hang until the slowly starting instance is open.
  2. Bug 26626720 - ASM rebalance may be very slow or appear to hang. 
  3. In both cases, wait stack typically shows processes waiting for "GCS lock open S".

Workaround

N/A

NOTE: The recommended resolution is to apply the patch.  As a temporary workaround, it is possible to limit the number of loopback IP addresses to one by setting an underscore initialization parameter.  However, configuring this workaround will prevent a customer from allocating more than 64k database processes, which may substantially impact a customer's ability to run normal workload. In such cases the patch must be applied to permit more than 64k processes.  Before configuring the workaround check v$process during peak workload to determine the number of processes the customer's application(s) must support.

To implement the workaround, set initialization parameter _bug20480623_posting_ips="127.0.0.1" in both ASM and database initialization files.  Ensure this parameter is removed after applying the patch.

Patches

Update grid infrastructure and database homes to 12.1.0.2.171017, or later.

Alternatively, apply <patch 26762083> to 12.1.0.2 grid infrastructure and database homes that have the fix for bug 25256170 installed.  Note that the patch available for 12.1.0.2.170718 is applicable to both 12.1.0.2.170718 and 12.1.0.2.170814 software homes.

 

History

18-Oct-2017 - Fixed in 12.1.0.2.171017; patch for 12.1.0.2.170718 also applicable to 12.1.0.2.170814
20-Sep-2017 - Published 


Attachments
This solution has no attachment
  Copyright © 2018 Oracle, Inc.  All rights reserved.
 Feedback