review list

devops
springboot and microservice
sgmlparser, design pattern
persistable queue, java volatile, atomacity and concurrency,  mockito


volatile is not atomic, happen-before, happen-after for memory visibility
AtomicLong methods compareAndSet, getAndIncrement, getAndSet, decrementAndGet, set, get, weakCompareAndSet
AtomicReference
await, signify,signifyall,interrupt, join, wait, sleep
thread state (new, runnbale, blocked, waiting, timed_waiting, terminated)
concurrentlinkedQueue is non-blocking queue. iterator is weak consistent. size method needs to tranverse entire collection and not accurate.
     head and tail. head doesn't store data. poll element from head, offer element to tail. node is deleted if its next points to itself.
concurrenthashmap, 
    size and containsValue needs to cal modCount multiple times and then tranverse all segments if necessary
    get, no lock is needed
    segment array is final. acquire and release locks in order to avoid deak-lock
   in java8, use Node+CAS+synchronized


hashmap resize when size > capacity*loadfactor. h & (length - 1). In java 8, if size of entry list is >= 7, transform entry list to red-black tree


sgmlparser, strategy pattern (sgml, xml, with or without dtd, dtd parser), state pattern(state transfer, actions for node types), decorator pattern (content handler, syntax validator, rule execution;)
configurable for entity resolver and node validator
syntax validator, ElementDecl, ElementDeclGroup. stacks
                  ParseContext to record file and line number in thread local
                   multiple threads, one for reader, one for consumer. exchanger to exchange char list. threads coordination, uncaughtexception handler,
                                              end char(-1) as killer pill, interrupt, waiting for dying flag, join.




python automation test


hadoop flow, map, partition, sort, combine, shuffle, group , reduce, yarn, configurations
spark, RDD and memory. request resource from yarn, spark code in executor，spark tuning
hbase, regions， security


devops (git, docker, CI, TDD)


go through the favorite


csdn blogs (sort algorithms and their use case, scala class initialization order, scala advanced topics)


linux auto-start services, fstable, cluster setup (authorized keys)




elasticsearch, sparksql,dataframes


java 8 fork/join, divide and consquer. work stealing feature compared to thread pool. recursiveaction, recursivetask, invokeall(task1, task2) and task1.join and task2.join


hbase incremental publish. multiple start and end keys


-------------
region split,
salted row prefix change
compression