Torstan

[置顶] select m objects from n objects randomly

摘要： Q: how to select m objects randomly from n objects with euqal possibility?A: allocate an array of m elements to keep the final result. put the first m... 阅读全文

posted @ 2013-12-24 21:57 Torstan 阅读(318) 评论(0) 推荐(0) 编辑

[置顶] How does gdb work?

摘要： http://www.alexonlinux.com/how-debugger-worksSIGTRAP is used as a mechanism for a debugger to be notified when theprocess it's debugging hits a breakp... 阅读全文

posted @ 2012-11-12 14:36 Torstan 阅读(287) 评论(0) 推荐(0) 编辑

[置顶] mathematical induction and well-ordering principle

摘要： 1. Mathematical inductionMathematical induction can be used to prove the statement that asserts P(n) is true for all positive integers n, where P(n) i... 阅读全文

posted @ 2012-09-07 20:20 Torstan 阅读(723) 评论(0) 推荐(0) 编辑

[置顶] understanding assembly code (1)

摘要： 1) BackgroundSometimes, when we meet with a problem very hard to explain or understand, we can try more powerful tools, e.g. math or assembly code. In... 阅读全文

posted @ 2012-07-22 21:18 Torstan 阅读(563) 评论(0) 推荐(0) 编辑

[置顶] introduction to matheatic logic

摘要： key conceptsproposition: a statement that is true of falsepropostional variable: a variable that represents a proposition, e.g., use p describe propos... 阅读全文

posted @ 2011-11-19 23:51 Torstan 阅读(464) 评论(0) 推荐(0) 编辑

gnu cc扩展和ABI

摘要： gnc cc扩展标准c的语法，非常强大！！！详情请见： http://uw714doc.sco.com/cgi-bin/info2html?%28gcc.info%29C%2520Extensions&lang=en http://tigcc.ticalc.org/doc/gnuexts.html# 阅读全文

posted @ 2017-03-11 19:01 Torstan 阅读(464) 评论(0) 推荐(0) 编辑

segv & mini coredump 调研

摘要： 1. mini coredump a. segv http://zh.scribd.com/doc/3726406/Crash-N-Burn-Writing-Linux-application-fault-handlers https://github.com/gby/libcrash b. /pr 阅读全文

posted @ 2016-01-15 22:42 Torstan 阅读(765) 评论(0) 推荐(0) 编辑

多线程+fork 引发的bug查找

摘要： 1. 问题描述某个server SA是一个多线程服务器，主线程会调用fork，再exec生成工作进程SB。实际上，SA的主线程fork出了一个子线程，但没有执行exec。# ps ajxf | grep r2server14022 28342 28341 14022 pts/2 28341 S+... 阅读全文

posted @ 2015-10-27 21:48 Torstan 阅读(616) 评论(0) 推荐(0) 编辑

简单对比Spark和Storm

摘要： 2013年参与开发了一个类似storm的自研系统， 2014年使用过spark 4个多月，对这两个系统都有一些了解。下面是我关于这两个系统的简单对比： Spark： 1. 基于数据并行，https://en.wikipedia.org/wiki/Data_parallelism。相同的操作作用在数阅读全文

posted @ 2015-07-14 22:18 Torstan 阅读(2238) 评论(0) 推荐(0) 编辑

notes: the architecture of GDB

摘要： 1. gdb structureat the largest scale,GDB can be said to have two sides to it:1. The "symbol side" is concerned with symbolic information about the pro... 阅读全文

posted @ 2015-03-04 23:09 Torstan 阅读(426) 评论(0) 推荐(0) 编辑

How systems researchers build systems

摘要： Define the problem>>Identify the constraints and abstract problempropose solution:simple idea>>Don't try to solve everythingimplent the system:>>reuse... 阅读全文

posted @ 2014-12-15 15:39 Torstan 阅读(179) 评论(0) 推荐(0) 编辑

spark1.1.0源码阅读-executor

摘要： 1. executor上执行launchTask1 def launchTask(2 context: ExecutorBackend, taskId: Long, taskName: String, serializedTask: ByteBuffer) {3 val tr... 阅读全文

posted @ 2014-12-11 22:36 Torstan 阅读(301) 评论(0) 推荐(0) 编辑

spark1.1.0源码阅读-taskScheduler

摘要： 1. sparkContext中设置createTaskScheduler 1 case "yarn-standalone" | "yarn-cluster" => 2 if (master == "yarn-standalone") { 3 logW... 阅读全文

posted @ 2014-12-11 22:28 Torstan 阅读(369) 评论(0) 推荐(0) 编辑

spark1.1.0源码阅读-dagscheduler and stage

摘要： 1. rdd action ->sparkContext.runJob->dagscheduler.runJob 1 def runJob[T, U: ClassTag]( 2 rdd: RDD[T], 3 func: (TaskContext, Iterator[T])... 阅读全文

posted @ 2014-12-11 20:46 Torstan 阅读(308) 评论(0) 推荐(0) 编辑

akka简单示例-2

摘要：手动敲了一遍计算pi的示例：http://www.gtan.com/akka_doc/intro/getting-started-first-scala.html有个笔误，花了半个小时定位。 1 [Torstani@sparkb5-i ~/akka_example/pi]$ cat src/Pi.s... 阅读全文

posted @ 2014-12-10 22:09 Torstan 阅读(338) 评论(0) 推荐(0) 编辑