How to Collect Errorstacks for use in Diagnosing Performance Issues. (文档 ID 1364257.1)

In this Document
  Goal
  Solution
     Introduction
     Gathering Errorstacks
     Gathering Errorstacks on Your Current session
     Command Explanation


APPLIES TO:

Oracle Server - Enterprise Edition - Version: 8.0.3.0 to 11.2.0.3 - Release: 8.0.3 to 11.2
Information in this document applies to any platform.

GOAL

This article outlines how to collect Errorstack trace for use in diagnosing performance issues.
Errorstack is an internal event and should only be set on the advice of Oracle Support.

SOLUTION

Introduction

The commands shown below provide the most useful set of commands to capture session related performance data for various problems under the guidance of Oracle Support.

Gathering Errorstacks

To dump trace and errorstacks, either the Operating system process id or Oracle process id for a slave process must be determined.
Assuming the Oracle SID for the process is known then the following select can be
used to find the Operating system process id :

SELECT p.pid, p.SPID,s.SID
FROM v$process p, v$session s
WHERE s.paddr = p.addr
AND s.SID = &SID;

SPID is the operating system identifier
SID is the Oracle session identifier
PID is the Oracle process identifier

and then do the following:

Lets assume that the slave process id to be dumped has an o/s pid of 9834
and an Oracle pid of 34

login to SQL*Plus:

connect / as sysdba
ALTER SESSION SET tracefile_identifier = 'STACK_10046';
oradebug setospid 9834
oradebug unlimit

oradebug event 10046 trace name context forever,level 12

oradebug dump errorstack 3
oradebug dump errorstack 3
oradebug dump errorstack 3
oradebug tracefile_name

To us the Oracle Pid instead, replace the :

oradebug setospid 9834

with

oradebug setorapid 34

Remember to change the PIDs to the actual values on your system!

The final line outputs the trace file name which will include the string 'STACK_10046' for easy identification.
To disable the tracing once tracing is finished:

oradebug event 10046 trace name context off

This will produce a trace file in the relevant trace destination matching the process traced.

Gathering Errorstacks on Your Current session

If you are trying to gather stacks on your current session, you can determine the PID values by running selects such as the following in that session:

SELECT p.pid, p.SPID,s.SID
FROM v$process p, v$session s
WHERE s.paddr = p.addr
AND s.audsid = userenv('SESSIONID') ;

or

SELECT p.pid, p.SPID,s.SID
FROM v$process p,v$session s
WHERE s.paddr = p.addr
AND s.SID =
(SELECT DISTINCT SID
FROM V$MYSTAT);

Command Explanation

The following is a brief outline of the purpose of the commands:
  • connect / as sysdba
    This provides the simplest way of gathering the required access to dump the information.

  • ALTER SESSION SET tracefile_identifier = 'STACK_10046';
    Adds a search-able string to the trace file for easy identification later.

  • oradebug setospid XXXX
    Connects the oradebug tool to a particular O/S (operating system) PID (Process ID)

  • oradebug setorapid XXXX
    Connects the oradebug tool to a particular oracle PID (Process ID)

  • oradebug unlimit
    Unrestricts the trace file size so as not to lose valuable diagnostics.

  • oradebug event 10046 trace name context forever,level 12
    Collects 10046 trace showing waits and binds. This trace is ofter useful when collected in conjunction with errorstacks and providing context. This is optional.

  • oradebug dump errorstack 3
    Dumps the process stack and process state for the current process.

  • oradebug tracefile_name
    Writes the identifier to the trace file.



深度学习是机器学习的一个子领域,它基于人工神经网络的研究,特别是利用多层次的神经网络来进行学习和模式识别。深度学习模型能够学习数据的高层次特征,这些特征对于图像和语音识别、自然语言处理、医学图像分析等应用至关重要。以下是深度学习的一些关键概念和组成部分: 1. **神经网络(Neural Networks)**:深度学习的基础是人工神经网络,它是由多个层组成的网络结构,包括输入层、隐藏层和输出层。每个层由多个神经元组成,神经元之间通过权重连接。 2. **前馈神经网络(Feedforward Neural Networks)**:这是最常见的神经网络类型,信息从输入层流向隐藏层,最终到达输出层。 3. **卷积神经网络(Convolutional Neural Networks, CNNs)**:这种网络特别适合处理具有网格结构的数据,如图像。它们使用卷积层来提取图像的特征。 4. **循环神经网络(Recurrent Neural Networks, RNNs)**:这种网络能够处理序列数据,如时间序列或自然语言,因为它们具有记忆功能,能够捕捉数据中的时间依赖性。 5. **长短期记忆网络(Long Short-Term Memory, LSTM)**:LSTM 是一种特殊的 RNN,它能够学习长期依赖关系,非常适合复杂的序列预测任务。 6. **生成对抗网络(Generative Adversarial Networks, GANs)**:由两个网络组成,一个生成器和一个判别器,它们相互竞争,生成器生成数据,判别器评估数据的真实性。 7. **深度学习框架**:如 TensorFlow、Keras、PyTorch 等,这些框架提供了构建、训练和部署深度学习模型的工具和库。 8. **激活函数(Activation Functions)**:如 ReLU、Sigmoid、Tanh 等,它们在神经网络中用于添加非线性,使得网络能够学习复杂的函数。 9. **损失函数(Loss Functions)**:用于评估模型的预测与真实值之间的差异,常见的损失函数包括均方误差(MSE)、交叉熵(Cross-Entropy)等。 10. **优化算法(Optimization Algorithms)**:如梯度下降(Gradient Descent)、随机梯度下降(SGD)、Adam 等,用于更新网络权重,以最小化损失函数。 11. **正则化(Regularization)**:技术如 Dropout、L1/L2 正则化等,用于防止模型过拟合。 12. **迁移学习(Transfer Learning)**:利用在一个任务上训练好的模型来提高另一个相关任务的性能。 深度学习在许多领域都取得了显著的成就,但它也面临着一些挑战,如对大量数据的依赖、模型的解释性差、计算资源消耗大等。研究人员正在不断探索新的方法来解决这些问题。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值