Martin's Blog » NSM

Fiddler Timers

Martin Liu — Wed, 22 Dec 2010 07:28:45 +0000

Fiddler is a Web Debugging Proxy which logs all HTTP(S) traffic between your computer and the Internet. Fiddler allows you to inspect all HTTP(S) traffic, set breakpoints, and “fiddle” with incoming or outgoing data. Fiddler includes a powerful event-based scripting subsystem, and can be extended using any .NET language.Fiddler is freeware and can debug traffic from virtually any application, including Internet Explorer, Mozilla Firefox, Opera, and thousands more.

ClientConnected- Exact time that the client browser made a TCP/IP connection to Fiddler.
ClientBeginRequest– Time at which this HTTP request began. May be much later than ClientConnected due to client connection reuse.
ClientDoneRequest – Exact time that the client browser finished sending the HTTP request to Fiddler.
DNSTime – # milliseconds Fiddler spent in DNS looking up the server’s IP address.
GatewayDeterminationTime – # milliseconds Fiddler spent determining the upstream gateway proxy to use (e.g. processing autoproxy script). Mutually exclusive to DNSTime.
TCPConnectTime – # milliseconds Fiddler spent TCP/IP connecting to that server’s IP address.
HTTPSHandshakeTime – Amount of time spent in HTTPS handshake
ServerConnected – Time at which this connection to the server was made. May be much earlier than ClientConnected due to server connection reuse.
FiddlerBeginRequest – The time at which Fiddler began sending the HTTP request to the server.
ServerGotRequest – Exact time that Fiddler finished (re)sending the HTTP request to the server.
ServerBeginResponse – Exact time that Fiddler got the first bytes of the server’s HTTP response.
ServerDoneResponse – Exact time that Fiddler got the last bytes of the server’s HTTP response.
ClientBeginResponse – Exact time that Fiddler began transmitting the HTTP response to the client browser.
ClientDoneResponse- Exact time that Fiddler finished transmitting the HTTP response to the client browser.

Can i find out how long the server needs to process my request ?
(ServerBeginResponse – ServerGotRequest) is probably what you want.

Can i find out how long the request needs to pass through the network?
I’m not sure that’s what you really want to know, but it would be:

(ServerGotRequest – ClientDoneRequest) – (DNSTime + TCPConnectTime)

No related posts.

Make Cacti more easier

Martin Liu — Fri, 23 Jul 2010 01:46:33 +0000

Cacti是我最喜爱的一个网管软件之一。网站http://cactiez.cactiusers.org/ ；该网站的主人吧Cacti嵌入到了Linux的安装光盘中，实现了Cacti的一键式安装，这正是我所关注的关于开源应用的重要的一个步骤。我称之这是在解决开源软件应用的“last mile”问题。开源软件的潜在用户往往有这样几个特点：

技术力量弱，有些可能根本就不知道什么是Linux，什么是open source
需求相对明确和简单
无法获得中文的技术支持和培训

CactiEz让人能在30分钟之内上手开始使用Cacti，从根本上解决了整套软件的安装和配置工作。

英文版下载地址：http://cactiez.cactiusers.org/

中文版下载地址：http://linux.chinaunix.net/bbs/thread-1049886-1-1.html

OpenNMS 1.6.1 to 1.6.10

Martin Liu — Sun, 28 Mar 2010 05:37:04 +0000

在最近的一年里，我基本上忙于CMDB的工作，没有什么业余时间来看看OpenNMS的情况。其实OpenNMS是最初引入我深入了解开源软件世界的东东。时隔1年之久，我上一篇关于opennms的文章是它是1.6.1版本，刚才查了一下，它最新的稳定版本是1.6.10.

看似它这一年里发展的比较慢，这让我感到些许地失望，大概地看一下一堆相关的release notes。功能性的变化如下：

新增和增强了一些monitor的和功能，主要有http，dns，ldap，ssh，jmx等，从最初的snmp采集，都这些采集功能；采集能力的增强是它这样的无代理监控程序永远的话题
UI的功能的增强了一些
SNMP采集仍然是他的主要能力，在这方面也有了一些改进
新增了一些网络设备品牌的支持
阀值配置也有改进

总的来说，作为一款能够采集各种snmp mib信息，能够出好看的性能报表，能够作为统一的告警事件平台，能够有简单易用的UI，OpenNMS做的显然是非常不错的。至今还行它还没有提出什么云计算相关的话题，可见他们还是一般比较稳健发展，不爱赶时髦的人。希望他们能做的更好。

对于我的这个blog来说，之前写的OpenNMS和网管相关的东西比较多，以后可能会越来越少，就此和OpenNMS做一个总结。以后本blog可能会和目前做的cmdb和itsm相关这些内容为主了。

Some useful tools

Martin Liu — Sat, 04 Jul 2009 05:41:06 +0000

I am just cleaning up my bookmarks. Some of them could not be deleted. I’d like to list here.

This is a netflow monitor tool, you can capture, graph and store data.

Founded in 1992, the Distributed Management Task Force, Inc. (DMTF) is the industry organization leading the development of management standards and integration technology for enterprise and Internet environments.

Social Networking Meets Business, Concursive have CRM, online communitie, collaboration tools.

No related posts.

[網播] Born to MISrepresent 第二集：當 OpenNMS 遇上 SFLC ，催生了 Moglen Ravicher LLC

Martin Liu — Wed, 04 Feb 2009 04:52:02 +0000

（原文於 2008-04-08 發表於 http://blog.roodo.com/ystuan/archives/5830995.html）

這個音檔的授權是姓名標示-非商業性-相同方式分享 2.5 台灣，跟我的網誌一樣。

下載ogg格式：Born_to_MISrepresent_EP2.ogg
下載mp3格式：Born_to_MISrepresent_EP2.mp3
前往存放在 Archive.org 上的頁面：Born to MISrepresent EP2，有更多資訊以及格式。

內容摘要：

OpenNMS 1.3.11 版發佈，並宣佈與 Hyperic 就雙方的產品進行合作。這項合作的推手是一個雙方的共同客戶。Hyperic 的 agent 預料將可和擁有強大平台但是拙於 agent (agent-less)的 OpenNMS 形成良好互補。（自 1.3.10 版就開始跟 Hyperic Agent進行整合，最近 OpenNMS 版本號已跳升至 1.5.90 ） Tarus 的 blog 有更多細節：The Year of Integration

Hyperic 跟 OpenNMS 的初步整合，可線上收看 Integrating Hyperic HQ and OpenNMS（瀏覽器需支援 flash 格式播放）

當 OpenNMS 遇上 SFLC ：前一陣子 Tarus 跑去 Ask Slashdot ，嚷嚷說 Cittio Watchtower 使用 OpenNMS 的程式碼當然可以，但是使用的方式違反了（程式碼採用的）GPL授權。

想也知道，跑去 Slashdot 一定會引來關注，他也尋求軟體自由法律中心（Software Freedom Law Center，SFLC）的協助。但是，SFLC服務對象僅限於非營利機構，結果呢，SFLC決定以成立一間新的律師事務所(Moglen Ravicher LLC)的方式，來服務支持自由軟體的營利機構， OpenNMS Group 因此成了 Moglen Ravicher LLC 的第一個客戶。

喔，不用猜， SFLC 的新聞稿直接告訴大家，兩塊招牌底下都是同一群律師，事務所的所有利潤會回饋給 SFLC 。（嗯，其實我們也可以說 OpenNMS 有兩種，一種是 opennms.org – 社群，一種則是 opennms.com – 主要參與者所開的公司 OpenNMS Group。）

筆者已經在 FLOSS Weekly 聽過 Eben Moglen 談 GPL 3.0 ，這次為多瞭解另一位大咖 Dan Ravicher （其實只是想搞清楚他的姓怎麼唸……），找到了 Dan 談 GPL 3, Patents and Other Current Issues ，都很棒，推薦給大家，尤其是後者，其實 Dan 還回顧了 GPL 的誕生，解釋了 RHEL 的 subscription model 為何不違反 GPL 但又可以約束客戶（超意外收穫！解答了筆者幾百年前的疑惑）

照例 Tarus 也在他的 blog 提到整件事：OpenNMS, Eben Moglen and Cittio

opennms 1.6.1-1 quick start guide

Martin Liu — Sun, 11 Jan 2009 02:33:12 +0000

本安装手册为纯命令版，如果有什么疑问清参考官方安装手册，或给本贴留言。

[install yum on your Linux box]
选择任意Linux系统，安装上yum工具，在命令行测试 yum -v

[testing internet connection with those two URL]
测试是否能上网，用浏览器分别打开这两个网址

先别关闭这两个网页。

[Install opennms 开始安装:

yum install yum-fastestmirror
rpm -Uvh http://yum.opennms.org/repofiles/opennms-repo-snapshot-rhel4.noarch.rpm
yum list opennms
yum install opennms

[ post-install and config] 开始配置：

export OPENNMS_HOME=/opt/opennms
vi /var/lib/pgsql/data/pg_hba.conf
1. 注释掉其中所有可用的行，加入下面三行
  local all all trust
  host all all 127.0.0.1/32 trust
  host all all ::1/128 trust
/sbin/service postgresql restart
/usr/java/jdk1.5.0_15/bin/java -version
$OPENNMS_HOME/bin/runjava -S /usr/java/jdk1.5.0_15/bin/java
$OPENNMS_HOME/bin/install -disU -l /usr/lib/jni:/usr/lib

[ startup opennms and login ]启动登录：

/etc/init.d/opennms start
http://localhost:8980/opennms
username and password are admin/admin
click Admin / Add Interface, input a ipaddress and press Add button

OK， I have get oepnnms 1.6.1-1 installed on my home pc, I hope you good luck and have fun.

[发布说明] OpenNMS 1.6.0 (Stable) Released

Martin Liu — Fri, 14 Nov 2008 14:57:36 +0000

OpenNMS 1.6.0 (Stable) Released

似乎每年的10月以后我就会进入一个超级繁忙的状态，很长时间没有更新blog了，今天姑且转载一个OpenNMS的发布说明吧。

OpenNMS, the world’s first enterprise-grade network management platform developed as 100% free and open software, has released version 1.6.0. This is a stable, production release that incorporates nearly three years of development.

以上算是OpenNMS的广告语大家看看就是，需要注意的是一个企业级的产品实施费用是从很低的门槛一直到企业级的花费的。和其他的企业级解决方案的实施没有区别，差异在于，一个不同的许可证类型，此为开源许可证是也。

The last production version, 1.2.0, was aimed to compete squarely with Hewlett-Packard’s OpenView Network Node Manager product. This release builds upon that work to expand the reach of OpenNMS to other parts of the OpenView family as well as to provide an open source alternative to products such as Tivoli’s Netcool.

以上是OpenNMS的精神所在：OpenNMS一直在叫嚣这和OpenView, NetCool较量；这足以说明了它的勇气，和发展方向，它的创始人Tarus从一个人单枪匹马开始，到现在有一般人inhouse开发这个产品，自己全世界出差，如果说没有实现和商业产品的抗衡的话，我个人觉得他起码实现了一个所有开源参与者的梦想“为需要自己的人工作，为自己工作”

OpenNMS 1.6.0 sports a redesigned user interface, a number of scalability improvements and increased integration with other products. OpenNMS now runs on Windows, in addition to most flavors of Linux, Solaris and Mac OS X.

上面说的是新版的用户界面重新设计了，目前我还没有时间安装，回头装好了一定上图片，新版的稳定性得到提高并增加了和其他产品的集成。现在OpenNMS也能在Windows上运行了，在我看这是早晚的是，它是个纯Java的应用，移植不是什么难事，不过我不推荐在windows上完开源的产品，有点不伦不类的感觉，而且从原生态的Linux环境中才哪呢个够获取开源的全部优势。

One of the major additions to the platform is the Alarms subsystem. OpenNMS can receive events from a number of sources, such as SNMP traps, syslog, TL/1, and custom scripts. A key can be configured for each event that will allow it to be turned into an alarm. Thus if a device is generating multiple, identical events, their number will be reduced into just a single alarm. This greatly reduces the amount of event “noise” that operators see.

上面说告警子系统现在是增强了。OpenNMS能收到像是snmp trap，syslog， TL/1和自定义脚本发来的告警事件。某种键值能被设置，让一个事件能被出发成一个警报。报警事件经过了压制和降噪处理，较少了重复报的数量。

In addition, automated actions can be performed on alarms. For example, events that signal problem resolution, or “up” alarms, can be matched with “down” alarms to automatically clear them. Event workflow can be built into the system by using these automations to manage the alarm list, thus freeing up the operators to focus on the most important issues.

另外，对于一个警报来说，可以和一个自动化的动作配合，例如某些up事件能去清除对应的down事件。

While OpenNMS contains a robust automated discovery system, when managing tens of thousands of nodes it is often preferred to allow an external system to determine what OpenNMS is to monitor. Thus OpenNMS 1.6.0 contains a new “model importer” feature that allows node, interface and service information to be imported directly into the system using data in an XML format. One company uses this method to manage over 70,000 devices with a single instance of OpenNMS.

同时OpenNMS现在的自动发现系统非常强大哦，档管理一万个节点的时候，OpenNMS往往倾向于使用一个外部的系统来决定什么是需要OpenNMS监控的。在1.6中包括这样一个新功能‘模型导入’，他能把xml格式的节点和端口数据自己导入到OpenNMS系统中。曾经一个公司用这个方法管理这7万个节点，都在一套OpenNMS系统内，好家伙~~也就是说这个功能有了之后，你就可以不用一个节点的去发现了，往往有些机器发现的那会正好不在线，那就漏掉很多的机器了。
Data collection saw many improvements as well. With the proper hardware, OpenNMS is able to collect over one million data points every five minutes. This data can be from SNMP (versions 1, 2c and 3), JMX, HTTP, or NSClient. The collected data can be exported via the web user interface. Reports showing the highest and lowest values for a particular set of data points (Top N Reports) can also be created, and 1.6.0 contains a vastly improved thresholding system. Thresholds can be generated on individual data points, combinations of data points, as well as a “relative change” such as when a value shows a sudden increase or decrease.

数据的采集方面也有很多增强。利用一个合适的硬件，OpenNMS能够每5分钟把超过一百万的数据采集点采集一遍，要是采集稳定的话，这还是比较快的了。被采集的数据点可以使来自 snmp v1,2,3, JMX, http, nsclient。采集来的数据能通过web界面到处了。这还是不错的功能，能有可能做一些系统之外的自定义报表了。自带的报表功能能够定制TopN的报表。OpenNMS现在的阀值管理系统也得到了增强。能为某一个数据点设置，也能为一组数据采集点，同时有了对 “relative change”的管理，也就是某个数据忽高忽低的管理，也即是某些瞬间的异常增的很高或降的很低。

OpenNMS was originally designed for network service monitoring, and that functionality has been increased as well. New monitors for such things as Windows services are now available, as well as more advanced synthetic transactions. The Page Sequence Monitor was created to monitor a complete web-based transaction, while the Mail Transport Monitor determines the full round-trip availability of a mail service.

OpenNMS天生是监控网络的。新增了很多monitor，例如windows服务监控，模拟交易监控。页面序列监控器用了监控一些列的web交易处理行为，还能通过Mail Transport Monitor 来确定邮件服务的可用性。

Probably the biggest change was the development of distributed monitoring. Using a small Java webstart application installed on a remote system, OpenNMS is able to monitor service availability from the point of view of the remote system. Combined with the Page Sequence Monitor one can measure the user’s experience when visiting a website from various remote locations.

最大的变更在这个版本中是‘分布式监控’，使用一个小的安装在远处的一个主机上的Java Webstat应用，opennms能过过这个监控服务点看到远程的系统上采集的数据。结合页面序列监控功能，它可以实现这样的功能：在南北方，网通电信的网络中找四个点安装这个程序后，模拟从不同网络位置监控某个web网站上一系列功能页面的可用性。这就是所谓的模拟交易管理，是Web应用管理方面中当下不可缺少的环节。

As OpenNMS was designed as a platform, there are numerous ways for external applications, both open and proprietary, to integrate with it. There is a new Trouble Ticketing API that allows for two-way communication between OpenNMS and a number of external ticketing systems such as Jira, Concursive (CentricCRM) and OTRS.
像opennms这样的平台，可以提供给开源或者商业外部系统的接口。通过这些api的双向接口，你可以把报警事件送到外部的工单管理系统(事件管理流程平台)中，例如 Jira, Concursive (CentricCRM) and OTRS.、

These are just a few of the new features available in 1.6.0. As always, OpenNMS is 100% free and open software. Please check it out and let us know what you think. We hope you enjoy using it as much as we enjoy creating it.

嗯，终于读完了整篇的发布说明，你能相信，它都发布了这么久了，我只有在周五晚上抽空仔细读完么。这周简直是太忙，天天加班，还有天天处理不完的状况。今天的一个教训就是“软件系统之间的集成不是简单能搞定的，需要小心，小心在小心”

NeDi – A Swiss Army Knife of network device management

Martin Liu — Sat, 13 Sep 2008 15:47:28 +0000

There is a shot list of its feature

Auto discovery network deivce
Network Topology-Map
SNMP based monitoring
Network performance and inventory management

NeDi movie will show you what it can do.

No related posts.

Systems Monitoring Shootout

Martin Liu — Mon, 04 Aug 2008 09:26:57 +0000

Please download it here A paper from Open Management Consortium

This is a paper from “2008 Ottawa Linux Symposium”. It will give you a nice insight about some great NSM projects. It is talking about OpenNMS, Zabix, Zenoss, GroundWorks and Hyperic, those might b the hottest projects around NMS field. If you are looking for a open source network and system monitoring solution, or you are testing one of them; you should check it out. I got this paper from Open Management Consortium.

Sometimes, people might spend too much time on testing different projects; I know this is a kind of fun. But eventually they only got lots of comments about manny project. They still did not realize the value of open source. The best way to adpat open source is that you just pick one nice project and keeping to use it for months at least. I have a net admin frind who I had since helped to setup Cacti for all of his network devices. He don’t know so many NSM projects, but he really engoy Cacti. With Cacti, he can do a easier and better job then before. So, are you going to still watching open source world? Let’s get start your real open source journey.

盛夏严防服务器中暑

Martin Liu — Sun, 15 Jun 2008 15:50:31 +0000

哦不是中暑:-)而是发烧；夏天就要到了，监控服务器的主板温度和风扇是否工作正常有变的比较重要起来了。

今天发现一个文档 Monitoring Temperature and Fan Speed Using Ganglia and Winbond Chips 这个是一个不错的文档，值得参考一下。

文章是写如何在Ganglia中实现对主板温度和风扇的监控。以前也有人问过我这个问题“Ganglia的监控图上为什么没有温度也风扇的图”；相信您看后就知道怎么回事了。

下面是我对关于温度和风扇监控的一点想法：

主板的芯片能提供这些数据的访问给操作系统
操作系统上有sensors这个命令工具用来采集这些数据
有合适的脚本分析上面命令的输出数据转换成能被监控工具（ganglia， cacti，nagios， opennms）采集的格式等
往往这两个参数并不是监控工具的默认采集数据，所以需要扩展采集工具的采集集合，收集并保存这些数据。
在监控工具如Ganglia中显示这些数据，当然如果有自动报警功能就好了，可以及时通知管理员给服务器消暑：）

Ganglia是最好的网格或者群集的监控软件，不过当你用它的使用也要注意到下面这个问题：

What does Ganglia not provide?
Ganglia does not attempt to address service monitoring or reporting (unlike Nagios). So far, we have not come across a single monitoring solution that addresses all of our needs effectively.

上面所说的服务监控是说，它的特点不是想Nagios和OpenNMS那样对服务器系统做非常全面细致的监控，它默认的监控指标比较少不过对于监控网格或者群集这种数量巨大的对象来说这些比较经典的指标也够用了，特别是对服务器的可用性，CUP、RAM，网络资源利用率和工作负载等数据的采集也比较够用了。如果你使用它来监控数量不是很多服务器的话，你可以使用它的可扩展性加入任何想监控的KPI。我最喜欢的还是它能把服务做分组，而且垂直的方向上可以加N层的嵌套；这种组织方式能比较好的适用于业务系统众多而复杂的数据中心。

Using Ganglia for cluster and grid monitoring