Environment: Database: 64-bit Oracle10.2.0.1.0TSM5.4: Windows2003 tape library: IBM3100 Problem description: 1. the aix system has a 3D32B80D error.
Environment: Database: 64-bit Oracle10.2.0.1.0 TSM5.4: Windows2003 tape library: IBM3100 Problem description: 1. the aix system has a 3D32B80D error.
Environment:
Database: 64-bit Oracle10.2.0.1.0 under AIX5.3
TSM5.4: Windows2003
With database: IBM3100
Problem description:
1. The error 3D32B80D occurs in the aix system. As follows:
JTXXDBS01:> errpt | more
Identifier timestamp t c RESOURCE_NAME DESCRIPTION
3D32B80D 0203175210 p s topsvcs NIM thread blocked
3D32B80D 0203175110 p s topsvcs NIM thread blocked
3D32B80D 0203175110 p s topsvcs NIM thread blocked
3D32B80D 0203175110 p s topsvcs NIM thread blocked
3D32B80D 0203174810 p s topsvcs NIM thread blocked
3D32B80D 0203174810 p s topsvcs NIM thread blocked
JTXXDBS01:> errpt-aj 3D32B80D | more
---------------------------------------------------------------------------
IDENTIFIER: 3D32B80D
Date/Time: Wed Feb 3 17:52:06 BEIST 2010
Sequence Number: 114902
Machine Id: 0005807BD600
Node Id: JTXXDBS01
Class: S
Type: PERM
Resource Name: topsvcs
Description
NIM thread blocked
Probable Causes
A thread in a Topology Services Network Interface Module (NIM) process
Was blocked
Topology Services NIM process cannot get timely access to CPU
User Causes
Excessive memory consumption is causing high memory contention
Excessive disk I/O is causing high memory contention
Recommended Actions
Examine I/O and memory activity on the system
Reduce load on the system
Tune virtual memory parameters
Call IBM Service if problem persists
Failure Causes
Excessive virtual memory activity prevents NIM from making progress
Excessive disk I/O traffic is interfering with paging I/O
Recommended Actions
Examine I/O and memory activity on the system
Reduce load on the system
Tune virtual memory parameters
Call IBM Service if problem persists
3D32B80D error is mainly caused by the high load of the entire system, resulting in HA heartbeat failure to communicate normally.
2. During backup, the disk IO frequency is mainly reflected in Hdisk0 and Hdisk1
We can see that the hard disk I/O is hdisk1 and hdisk0, and the busy disk I/O is nocomp (non-computing memory ).
Use lsps-s to view,
Let's look at it again:
Hdisk1 and hdisk0 physical volume AIX swap space. However, the usage of swap space is less than 30%.
This mainly shows that the paging space of AIX is sufficient, but the paging space is frequently exchanged. Note that the nocomp (non-computing memory) in the memory is up to 76%. The memory is not released for a long time, resulting in insufficient comp (computing memory) and frequent use of paging space.
3. NMON Monitoring Chart of AIX
The backup memory is seriously insufficient.
The paging space of AIX is frequently exchanged.
Specific reasons:
This is because the database is backed up by rman at every day, and disk I/O is frequent. The I/O of the AIX system is the input and output of the disk and the Paging Space respectively, the backup here is mainly reflected in the frequent operations of paging space. Among them, the non-computing memory reaches 76% and will not be released for a long time. The root cause is that non-computing memory is not released for a long time, and the AIX memory is missing, resulting in frequent Paging Space exchanges. As a result, the system load is relatively high.
Limit the maximum nocomp (non-computing) memory size of AIX.
Put the AIX
Minperm % = 20
Maxclient % = 80
Maxperm % = 80
Lower adjustment.
Solution:
View the minperm %, maxclient %, and maxperm % values of the original AIX. As follows:
Root User View:
> Vmo-
Minperm % = 20
Maxclient % = 80
Maxperm % = 80
Set minperm % to 10, maxclient % to 20, and maxperm % to 20
Run the following command with the root command line:
1,
Vmo-o minperm % = 10
2,
Vmo-o maxclient % = 20
3,
Vmo-o maxperm % = 20
If you change maxperm % to 20, the 3D32B80D error will not occur when you back up again.
Note:
Values of minperm and maxperm Parameters
The operating system satisfies different requirements by leaving the previously read/write memory pages in the real memory. If file pages are requested before their page frames are reassigned, the input and output operations are saved. The file page can be from a local or remote (such as NFS) system.
The ratio of the files used by PAGE frames to the files used for calculation (work or program text) is loosely controlled by the values of minperm and maxperm:
If the percentage of file pages in RAM is higher than maxperm, Page Replacement theft is only used for file pages.
If the percentage of file pages in RAM is lower than minperm, the theft of page replacement is also used for file pages and computing pages.
If the percentage of file pages in RAM is between minperm and maxperm, page replacement only steals file pages, unless the number of re-calls to the file page is greater than the calculated page number.
Noncomp is between 20% and 80%. In this case, the Memory Page switching algorithm only exchanges File Cache pages. However, if the swap rate of File Cache pages is greater than that of program pages, program pages are also exchanged. Because SGA is a Comp page and Data File Cache is a Noncomp page, when a large amount of data is accessed, the swap rate of the File Cache page is higher than that of the program page (SGA), so a large number of SGA pages are exchanged; to avoid switching a large number of SGA pages, you need to reduce the maxperm % value to less than 35%, so that only the File Cache page is exchanged, reducing the number of SGA exchanges.
5) Conclusion
Adjust the values of maxperm % and minperm % to set MINPERM and MAXPERM to 5%, 20%, or even smaller, so that the memory is more used in the SGA of Oracle than the file cache of the system.
Specific command: # vmo-o minperm % = 5-o maxperm % = 20
Note: You can add the-p parameter to permanently change the system configuration. Otherwise, only the current system configuration is changed and the System Configuration becomes invalid after the restart.