[20190211]简单测试端口是否打开.txt
–//昨天看一个链接,提到如果判断一个端口是否打开可以简单执行如下:
–//参考链接:https://dba010.com/2019/02/04/check-if-a-port-on-a-remote-system-is-reachable-without-telnet/
1.测试:
# cat < /dev/tcp/192.168.100.40/1521
# echo $?
0
# cat < /dev/tcp/192.168.100.40/1522
-bash: connect: connection refused
-bash: /dev/tcp/192.168.100.40/1522: connection refused
# alias zdate
alias zdate=’date +’\”%y/%m/%d %t’\”’
–//不过我的测试需要等待一分钟,不知道那个受那个内核参数的影响,不知道那位知道:
# zdate ;cat < /dev/tcp/192.168.100.40/1521 ; zdate
2019/02/11 08:42:59
2019/02/11 08:43:59
–//换另外一台机器测试:
# zdate ;cat < /dev/tcp/192.168.100.78/1521;zdate
2019/02/11 08:46:07
2019/02/11 08:46:17
# zdate ;cat < /dev/tcp/192.168.100.40/1521;zdate
2019/02/11 08:46:36
2019/02/11 08:47:36
–//视乎和远程服务器某个内核参数有关.测试100.40需要60秒(这台机器很老redhat 4.3的版本),而测试100.78需要10秒.
–//在100.78上使用tcpdump监测:
# tcpdump -i eth0 host 192.168.xxx.xx -nnn
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type en10mb (ethernet), capture size 96 bytes
09:03:41.242607 ip 192.168.xxx.xx.48292 > 192.168.100.78.1521: s 226276375:226276375(0) win 29200 <mss 1460,sackok,timestamp 3393435130 0,nop,wscale 7>
09:03:41.242785 ip 192.168.100.78.1521 > 192.168.xxx.xx.48292: s 995705425:995705425(0) ack 226276376 win 14480 <mss 1460,sackok,timestamp 1354661055 3393435130,nop,wscale 7>
09:03:41.242852 ip 192.168.xxx.xx.48292 > 192.168.100.78.1521: . ack 1 win 229 <nop,nop,timestamp 3393435130 1354661055>
09:03:51.253449 ip 192.168.100.78.1521 > 192.168.xxx.xx.48292: f 1:1(0) ack 1 win 114 <nop,nop,timestamp 1354671066 3393435130>
09:03:51.253770 ip 192.168.xxx.xx.48292 > 192.168.100.78.1521: f 1:1(0) ack 2 win 229 <nop,nop,timestamp 3393445141 1354671066>
09:03:51.253797 ip 192.168.100.78.1521 > 192.168.xxx.xx.48292: . ack 2 win 114 <nop,nop,timestamp 1354671066 3393445141>
–//也可以看出有10秒的等待时间.在100.78上执行:
# echo /proc/sys/net/ipv4/* | xargs -n 1 strings -1 -f | grep “: 10$”
/proc/sys/net/ipv4/cipso_cache_bucket_size: 10
/proc/sys/net/ipv4/igmp_max_msf: 10
/proc/sys/net/ipv4/inet_peer_gc_mintime: 10
/proc/sys/net/ipv4/tcp_keepalive_intvl: 10
*/
–//开始以为/proc/sys/net/ipv4/tcp_keepalive_intvl: 10有关,我查看100.40的设置/proc/sys/net/ipv4/tcp_keepalive_intvl=75,也不是这个
–//参数.我修改这个参数测试还是10秒.
–//实际上还有1个更快的测试方法,发送信息到/dev/tcp/ip_address/port,例子如下:
# echo a > /dev/tcp/192.168.100.78/1521
# echo $?
0
# echo a > /dev/tcp/192.168.100.78/1522
-bash: connect: 拒绝连接
-bash: /dev/tcp/192.168.100.78/1522: 拒绝连接
–//在100.78上使用tcpdump监测:
# tcpdump -i eth0 host 192.168.xxx.xx -nnn
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type en10mb (ethernet), capture size 96 bytes
09:12:41.312584 ip 192.168.xxx.xx.48331 > 192.168.100.78.1521: s 3309983125:3309983125(0) win 29200 <mss 1460,sackok,timestamp 3393975215 0,nop,wscale 7>
09:12:41.312761 ip 192.168.100.78.1521 > 192.168.xxx.xx.48331: s 124881852:124881852(0) ack 3309983126 win 14480 <mss 1460,sackok,timestamp 1355201125 3393975215,nop,wscale 7>
09:12:41.312850 ip 192.168.xxx.xx.48331 > 192.168.100.78.1521: . ack 1 win 229 <nop,nop,timestamp 3393975216 1355201125>
09:12:41.312895 ip 192.168.xxx.xx.48331 > 192.168.100.78.1521: p 1:3(2) ack 1 win 229 <nop,nop,timestamp 3393975216 1355201125>
09:12:41.312911 ip 192.168.100.78.1521 > 192.168.xxx.xx.48331: . ack 3 win 114 <nop,nop,timestamp 1355201125 3393975216>
09:12:41.312922 ip 192.168.xxx.xx.48331 > 192.168.100.78.1521: f 3:3(0) ack 1 win 229 <nop,nop,timestamp 3393975216 1355201125>
09:12:41.313454 ip 192.168.100.78.1521 > 192.168.xxx.xx.48331: f 1:1(0) ack 4 win 114 <nop,nop,timestamp 1355201126 3393975216>
09:12:41.313661 ip 192.168.xxx.xx.48331 > 192.168.100.78.1521: . ack 2 win 229 <nop,nop,timestamp 3393975217 1355201126>
–//这样几乎瞬间响应.后面2个网络包是监测1522端口的.
09:13:08.704102 ip 192.168.xxx.xx.15174 > 192.168.100.78.1522: s 2841139274:2841139274(0) win 29200 <mss 1460,sackok,timestamp 3394002608 0,nop,wscale 7>
09:13:08.704140 ip 192.168.100.78.1522 > 192.168.xxx.xx.15174: r 0:0(0) ack 2841139275 win 0
–//前面测试在root用户下进行,我在oracle用户下测试看看:
# zdate ;cat < /dev/tcp/192.168.100.40/1521 ; zdate
2019/02/11 09:22:51
2019/02/11 09:23:51
–//也是可以的.
2.继续测试:
–//我反复测试发现仅仅探测100.78需要10秒,其它的机器都是60秒.这样可能与内核的参数无关,因为我们服务器许多版本使用与100.78一样的.
–//我最终测试与参数inbound_connect_timeout_listener设置有关(在$oracle_home/network/admin/listener.ora下).
$ grep inbound_connect_timeout_listener listener.ora
inbound_connect_timeout_listener=10
–//而这个参数缺省就是60秒.修改它等于1,要重启监听看看(我测试要重启监听才生效):
# zdate ;cat < /dev/tcp/192.168.100.78/1521;zdate
2019/02/11 09:36:14
2019/02/11 09:36:15
–//ok现在需要1秒.
–//一般以前使用nc或者telnet判断,链接如下:
# echo a | nc -w 1 -n -v 192.168.xxx.xx 1-1521 2>/dev/null | grep “succeeded”
–//参考链接:http://blog.itpub.net/267265/viewspace-2148257/
–//关于inbound_connect_timeout的设置,在sqlnet.ora与listener.ora意义不一样,一个表示等待用户认证超时的时间,一个表示等待用
–//户连接请求超时的时间. 可以参考我以前链接:
–// http://blog.itpub.net/267265/viewspace-2147529/
3.自己在重复验证看看.
$ grep inbound_connect_timeout sqlnet.ora listener.ora
sqlnet.ora:sqlnet.inbound_connect_timeout=30
listener.ora:inbound_connect_timeout_listener=1
$ ps -ef | grep oracleboo[k]
–//没有任何输出.也就是远程没有客户连接服务器.
$ zdate;sqlplus scott/xxxxx@book
2019/02/11 09:48:04
sql*plus: release 11.2.0.4.0 production on mon feb 11 09:48:04 2019
copyright (c) 1982, 2013, oracle. all rights reserved.
error:
ora-01017: invalid username/password; logon denied
–//注意sqlplus不能使用-l参数,不然马上退出了.
–//切换终端会话,执行如下:
$ ps -ef | grep oracleboo[k]
oracle 43849 1 0 09:48 ? 00:00:00 oraclebook (local=no)
–//可以发现实际上网络连接已经建立.
–//等alert*.log出现如下信息,在取消上面sqlplus的执行:
–//alert.log
mon feb 11 09:48:34 2019
warning: inbound connection timed out (ora-3136)
–//两者时间相减就是30秒.这个参数也就是sqlnet.ora有关.也就是等待用户认证超时的时间.
$ zdate ;cat < /dev/tcp/192.168.100.78/1521;zdate
2019/02/11 10:00:21
2019/02/11 10:00:22
–//正好1秒,这个时候就是表示等待用户连接请求超时的时间.
$ zdate; telnet 192.168.100.78 1521;zdate
2019/02/11 10:01:38
trying 192.168.100.78…
connected to xxxxxdg4.com (192.168.100.78).
escape character is ‘^]’.
connection closed by foreign host.
2019/02/11 10:01:39
$ time telnet 192.168.100.78 1521
trying 192.168.100.78…
connected to xxxxxdg4.com (192.168.100.78).
escape character is ‘^]’.
connection closed by foreign host.
real 0m1.007s
user 0m0.001s
sys 0m0.003s
$ /usr/bin/time -p telnet 192.168.100.78 1521 > /dev/null
connection closed by foreign host.
command exited with non-zero status 1
real 1.00
user 0.00
sys 0.00