an efficient ssh toolkit 一个高效的工具，可以实现并行ssh连接到多台服务器进行操作，极大地提升效率。

##USAGE: nprsh [-sN] [-l] [-i] [-c|-C] [-T] [-rackprefix PREFIX -rackscale M.N] [-on NODE_LIST] [-f NODE_LIST_FILE] [-F NODE_TABLE [add |remove |-q QUERY_STRING] ] [-e EXCLUDED_NODE_LIST] [-x EXCLUDED_NODE_LIST_FILE] [-E EXCLUDED_NODE_TABLE [-q QUERY_STRING] ] <COMMAND |-b BATCH_FILE |-np NUM [-t] MPIAPP |-up |-L |-mpiv> ###TARGET SPECIFICATION: -rackprefix PREFIX -rackscale M.N Specify the valid node range, {PREFIX}{*}M..N. PREFIX defaults to "gs" Background: It is very common that nodes are not continually numbered in a cluster. The NEXT node of gs0149 may be gs0210, instead of gs0150. You can specify "-rackscale" as 10.49 to weed out gs0150..0209 from gs0110..0249. -on NODE_LIST Specify target nodes. NODE_LIST may be in the form of: 1. HOSTX 2. HOSTX..Y 3. G[ABC]HOSTX 4. G[ABC]HOSTX..Y 5. NODE_LIST,NODE_LIST 6. IP lists, such as 10.10.11.1-108 7. HOSTX..Y/N, the step is N If omitted, get target nodes from ~/.nprsh_nodes CAUTION: Valid characters in a node name are [A-Za-z0-9_-] -f NODE_LIST_FILE Specify NODE_LIST in a file. -F NODE_TABLE Specify NODE_TABLE to: 1. fetch nodes (also use -L), 2. update the table, You can use "-q" to add a database query string. -d "ND1D2" Specify that hostnames are extracted from the Nth field delimited by D1 and D2. -e EXCLUDED_NODE_LIST Nodes in EXCLUDED_NODE_LIST will not be used. Attention: Always get excluded nodes from ~/.nprsh_badnodes -x EXCLUDED_NODE_LIST_FILE Specify nodes that should be excluded. -E EXCLUDED_NODE_TABLE Specify EXCLUDED_NODE_TABLE. Query string WILL BE supported. -g PATTERN Specify the pattern to select dest nodes. Hint: You can specify two or more patterns using "|" -p PREFIX Specify the pattern to add before each node name. -I Ignore the list of badnodes. -X Equivalent to "-x /tmp/nprsh_chenzhenfeng.dnlst". ###FUNCTIONS: -A USER:PASSWD Execute as USER:PASSWD on target nodes. -b BATCH_FILE Execute COMMANDs listed in BATCH_FILE in BATCH mode. -sN Run on the target-nodes one by one, with N seconds of interval (at most 300 seconds). -l Run locally. Do not use in BATCH mode. In this mode, %DD will be substituted by a hostname! -L List all the target hostname and EXIT. -i Interactive mode. -D DT_DEGREE File distribution mode. -S STRIDE Stride in pairtest mode (-net). -U USER GROUP PASSWD Create USER in GROUP in batch mode. -P USER GROUP Make ssh pass-free on a list of hosts. -np NUM [-t] MPIAPP Run MPIAPP on cores. "-t", generate config files from a four-var template. -up Check up/down of each node. -rst Hard-reset a list of nodes (USE WITH CAUTION!!!). -kai Power ON a list of nodes (USE WITH CAUTION!!!). -guan Power OFF a list of nodes (USE WITH CAUTION!!!). -load Check the load of each node/cpu. -push Distribute and Run a script or a binary executable in the dest node(s). -net CMD ARGS Run pairtest between each pair of dest nodes. --install Local installation of nprsh for current user. -import user comments Create a host table, with a name in the style of 20100507_122503_hzg -kick reason_code NODE_LIST Put a list of nodes into in_repair state, the $default_htable, which is defined in NPRSHRC or using "-F", will be updated accordingly. -pull Reclaim a list of hosts after repair. -update Update the version of all available hosts in a host table. -checkt Check the current state of a list of hosts. -listtable USER [START [STOP]] List host tables created by USER, after START, and before STOP. -set field value Set the specified property of a list of host. ###MISC: -c Color off. -C Color on. -T Print the timestamp of each line. -q Be quiet. -V Be verbose/verboser. ###CAUTION: In parallel mode (default), the output from each target nodes will be messed up randomly. ###EXAMPLES: nprsh -on g[abc]node004..128 "hostname >/etc/HOSTNAME" ###init nprsh -on anode004..128 -e anode007,anode098 /etc/init.d/sshd start nprsh -f oldnodes -x badnodes -f newnodes date nprsh -on a118,a119,a130 -U ct test ct123 nprsh -on m1..32 -up nprsh -g a1|c19 -L nprsh -on [abc]110..159 -t -np 64 xhpl nprsh -on a110..159 scp /tmp/netdiag.log node1:/tmp/netdiag_%DD.log nprsh -A foo:foo123 CMD nprsh -on m1..32 -X -l scp nprsh.pl %DD:/usr/local/bin/nprsh nprsh -on m19,m28 -c "dmesg |tail" |vi - nprsh -on m1..32 -X -l scp %DD:/etc/rc.d/rc.local rc.local_%DD nprsh -on m1..32 -X -l scp slave_etc_hosts.%DD %DD:/slave/etc/hosts nprsh -on gh10..89 -X -l ssh %DD ifconfig ib0 %DD/g/i/ up