If you're familiar with UNIX, the pipeline, or pipe, is an integral part of everyday processing. Originally developed by Malcolm McIlroy, the pipeline allows you to redirect the standard output (stdout) of one command to become the standard input (stdin) of the following command in a single chained execution. Using the pipeline isn't limited to one instance per execution. Quite often, the stdout of one command is used as stdin of the following command, and the subsequent stdout is redirected yet again as stdin to another command and so on.
For example, one of the first things most UNIX administrators do on their systems during troubleshooting or daily checks is look at processes running currently on the system. Listing 1 shows such a check.
Listing 1. Example of a daily process check
# ps –ef
UID PID PPID C STIME TTY TIME CMD
root 1 0 0 Jul 27 - 0:05 /etc/init
root 53442 151674 0 Jul 27 - 0:00 /usr/sbin/syslogd
root 57426 1 0 Jul 27 - 0:00 /usr/lib/errdemon
root 61510 1 0 Jul 27 - 23:55 /usr/sbin/syncd 60
root 65634 1 0 Jul 27 - 0:00 /usr/ccs/bin/shlap64
root 82002 110652 0 Jul 27 - 0:24 /usr/lpp/X11/bin/X -x abx
-x dbe -x GLX -D /usr/lib/X11//rgb -T -force :0 -auth /var/dt/A:0-SfIdMa
root 86102 1 0 Jul 27 - 0:00 /usr/lib/methods/ssa_daemon -l ssa0
root 106538 151674 0 Jul 27 - 0:01 sendmail: accepting connections
root 110652 1 0 Jul 27 - 0:00 /usr/dt/bin/dtlogin -daemon
root 114754 118854 0 Jul 27 - 20:22 dtgreet
root 118854 110652 0 Jul 27 - 0:00 dtlogin <:0> -daemon
root 131088 1 0 Jul 27 - 0:07 /usr/atria/etc/lockmgr
-a /var/adm/atria/almd -q 1024 -u 256 -f 256
root 147584 1 0 Jul 27 - 0:01 /usr/sbin/cron
root 155816 151674 0 Jul 27 - 0:04 /usr/sbin/portmap
root 163968 151674 0 Jul 27 - 0:00 /usr/sbin/qdaemon
root 168018 151674 0 Jul 27 - 0:00 /usr/sbin/inetd
root 172116 151674 0 Jul 27 - 0:03 /usr/sbin/xntpd
root 180314 151674 0 Jul 27 - 0:19 /usr/sbin/snmpmibd
root 184414 151674 0 Jul 27 - 0:21 /usr/sbin/aixmibd
root 188512 151674 0 Jul 27 - 0:20 /usr/sbin/hostmibd
root 192608 151674 0 Jul 27 - 7:46 /usr/sbin/muxatmd
root 196718 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.mountd
root 200818 151674 0 Jul 27 - 0:00 /usr/sbin/biod 6
root 213108 151674 0 Jul 27 - 0:00 /usr/sbin/nfsd 3891
root 221304 245894 0 Jul 27 - 0:05 /bin/nsrexecd
daemon 225402 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.statd
root 229498 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.lockd
root 241794 151674 0 Jul 27 - 0:51 /usr/lib/netsvc/yp/ypbind
root 245894 1 0 Jul 27 - 0:00 /bin/nsrexecd
root 253960 1 0 Jul 27 - 0:00 ./mflm_manager
root 274568 151674 0 Jul 27 - 0:00 /usr/sbin/sshd -D
root 282766 1 0 Jul 27 lft0 0:00 /usr/sbin/getty /dev/console
root 290958 1 0 Jul 27 - 0:00 /usr/lpp/diagnostics/bin/diagd
root 315646 151674 0 Jul 27 - 0:00 /usr/sbin/lpd
root 319664 1 0 Jul 27 - 0:00 /usr/atria/etc/albd_server
root 340144 168018 0 12:34:56 - 0:00 rpc.ttdbserver 100083 1
root 376846 168018 0 Jul 30 - 0:00 rlogind
cormany 409708 569522 0 19:29:27 pts/1 0:00 -ksh
root 569522 168018 0 19:29:26 - 0:00 rlogind
cormany 733188 409708 3 19:30:34 pts/1 0:00 ps -ef
root 749668 168018 0 Jul 30 - 0:00 rlogind
|
The listing of the processes currently running on a system can be simple, as shown in Listing 1; however, most production systems run several more processes that make the output of
ps much longer. To shorten the list to what you're looking for, redirect the standard output of ps –ef using a pipeline to grep to search for exactly what you want to see. Listing 2 shows the process list from Listing 1 redirected to grep to search for the strings "rpc" and "ksh."Listing 2. Redirecting the process list to grep
# ps –ef | grep –E "rpc|ksh"
root 196718 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.mountd
daemon 225402 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.statd
root 229498 151674 0 11:00:27 - 0:00 /usr/sbin/rpc.lockd
root 340144 168018 0 12:34:56 - 0:00 rpc.ttdbserver 100083 1
cormany 409708 569522 0 19:29:27 pts/1 0:00 -ksh
cormany 733202 409708 0 19:52:20 pts/1 0:00 grep -E rpc|ksh
|
Using the pipeline can be much more complicated when you redirect stdout to stdin several times. In the following example, the previous
ps and grep example is expanded to pipeline the stdout to another grep to exclude any previous strings found that includes "grep" or "ttdbserver." When the final grep operation has finished, the stdout is redirected again using a pipeline to anawk statement to print any of the processes found with a process identifier (PID) larger than 200,000:# ps –ef | grep –E "rpc|ksh" | grep -vE "grep|rpc.ttdbserver" |
awk -v _MAX_PID=200000 '{if ($2 > _MAX_PID) {print "PID for
process",$8,"is greater than", _MAX_PID}}'
PID for process /usr/sbin/rpc.statd is greater than 200000
PID for process /usr/sbin/rpc.lockd is greater than 200000
PID for process -ksh is greater than 200000
|
Figure 1 provides a graphical representation of the command's stdout redirecting to stdin for the subsequent command.
Figure 1. Pipeline example
No comments:
Post a Comment