51Testing软件测试论坛

标题: loadrunner不知道如何监控内存 [打印本页]

作者: gabenmao 时间: 2010-11-23 14:14
标题: loadrunner不知道如何监控内存
安装rstatd守护进程后，已经监控到CPU和Averager load，但是Swap-in rate和Swap-out rate显示为0，不知道如何监控Linux内存？？
内存
Page-in rate
每秒钟读入到物理内存中的页数
Page-out rate
每秒钟写入页面文件和从物理内存中删除的页数
Paging rate
每秒钟读入物理内存或写入页面文件的页数，如果持续在几百，可能要加大内存了
这个要怎么分析？

作者: burenshi 时间: 2010-11-23 14:32

显示为零是说已经有这两个曲线，只是值一直为0吗？
如果是的话，就要跑点程序压一下服务器，这样才能看到曲线起伏，
如果不跑程序，只是连接到服务器，是会出现这种情况
不知道能帮到你不

作者: liujinkui 时间: 2010-11-23 14:46
试试这个
如果Process\Private Bytes计数器和Process\Working Set计数器的值持续升高
同时Memory\Available bytes计数器的值却持续降低的话
说明很有可能是存在内存泄漏

作者: gabenmao 时间: 2010-11-23 14:54
回复 2# burenshi

谢谢你~

作者: gabenmao 时间: 2010-11-23 14:55
回复 3# liujinkui

谢谢你~可是跑程序了，还是0，难道是linux有问题吗？
还有个问题就是：
内存
Page-in rate
每秒钟读入到物理内存中的页数
Page-out rate
每秒钟写入页面文件和从物理内存中删除的页数
Paging rate
每秒钟读入物理内存或写入页面文件的页数，如果持续在几百，可能要加大内存了

这三条如果有数值的话怎么判断内存的占用率呢？

作者: myallways 时间: 2010-11-23 20:21
使用Nmon监控吧

作者: xiaowu_627 时间: 2010-11-24 09:23
请问这个Nmon是独立于LoadRunner的一个工具吗？还是和LoadRunner有关系的啊？

作者: myallways 时间: 2010-11-24 15:48
独立的一个监控aix 和linux操作系统的一个小工具，很好用！！

作者: caiw0418 时间: 2010-11-24 15:58
回复 8# myallways

是什么能否写具体点能否发给我

作者: caiw0418 时间: 2010-11-24 16:05
回复 3# liujinkui

恩谢谢

作者: PrefTest 时间: 2010-11-24 18:45
Available Bytes Counter
Indicates the amount of physical memory available to processes running on the computer
A consistent value of less than 20–25% of installed RAM is an indication of insufficient memory.

Working Set Counter
If memory is scarce, Process(n)\Working Set tells you how much RAM each process is using.

Pages/sec Counter
Primary indicator to determine whether real memory is a potential bottleneck.

Page Reads/sec counters are hard page faults. A running thread has referenced a page in virtual memory that is not in the process working set. Nor is it a trimmed page marked in transition, but rather is still resident in memory. The thread is delayed for the duration of the I/O operation to fetch the page from disk. The operating system copies the page from disk to an available page in RAM and then redispatches the thread.

Watch out when Pages/sec exceeds 50 per paging disk.
Page Reads/sec Counter
Primary indicator to determine whether real memory is a potential bottleneck.

This counter indicates that the working set of the process is too large for the physical memory and that it is paging to disk. It shows the number of read operations, without regard to the number of pages retrieved in each operation. Higher values indicate a memory bottleneck.

If a low rate of page-read operations coincides with high values for Physical Disk\% Disk Time and Physical Disk\Avg. Disk Queue Length, there could be a disk bottleneck. If an increase in queue length is not accompanied by a decrease in the pages-read rate, a memory shortage exists.

Sustained values of more than five indicate a large number of page faults for read requests.

Pool Nonpaged Bytes Counter
Pages allocated from the Nonpaged pool are always resident in RAM.

Status information about every TCP connection is stored in the Nonpaged pool. Divide by the size of a page to calculate the number of allocated pages.

Watch the value of Memory\Pool Nonpaged Bytes for an increase of 10 percent or more from its value at system startup. If it indeed happens, a significant memory leak is in place.

Paged Pool Bytes Counter
Primarily used to identify processes that are leaking memory.

The number of committed virtual memory pages in the system’s Paged pool. System functions allocate virtual memory pages that are eligible to be paged out from the Paged pool. System functions that are called by processes also allocate virtual memory pages from the Paged pool.

Memory\Paged Pool Bytes reports how much virtual memory is allocated in the system Paged pool.Memory\Paged Pool Resident Bytes is the current number of Paged pool pages that are resident in RAM.The remainder is paged out.

Process(n)\Paged Pool Bytes increase of more than 10% for a specific process may point to leaking memory behavior.

Paged Pool Failures Counter
Server\Paged Pool Failures Counter:
Primarily used to identify a virtual memory shortage in the Paged pool.

The file Server service has a number of functions that allocate virtual memory pages from the Paged pool. If a memory leak exhausts the Paged pool, the file Server service might encounter difficulty in allocating virtual memory from the Paged pool. If a call to allocate virtual memory fails, the file Server service recovers gracefully from these failures and reports on them. Because many other applications and system functions do not recover gracefully from virtual memory allocation failures, this counter can be the only reliable indicator that a memory leak caused these allocation failures.

Any nonzero value of this counter indicates a bottleneck.

Committed Bytes Counter
The Committed Bytes:RAM ratio is a secondary indicator of a real memory shortage.

When Committed Bytes:RAM ratio exceeds 1.5, it clearly indicates real memory bottleneck.

作者: chase001 时间: 2010-11-24 23:46

试试这个
如果Process\Private Bytes计数器和Process\Working Set计数器的值持续升高
同时Memory\Availab ...
liujinkui 发表于 2010-11-23 14:46

版主可以来个详细讲解吗？学习下怎样判断内存泄漏

作者: caiw0418 时间: 2010-11-25 11:21
回复 12# chase001

试试内存泄露是怎么看

欢迎光临 51Testing软件测试论坛 (http://bbs.51testing.com/)