Linux performance monitoring commands

uptime --The amount of time the system up and running

last --Log of system shutdowns and changes in run level

ps -flu oracle:  Processes are owned by the user oracle

pidof ora_arc0_LABDB1:  process identifier

free, cat /proc/meminfo

ipcs –m:  Shared memory segments on the system

ipcrm –M : delete shared memory segments by a user

pmap –x 2103 : details memory mapped by that particular process

ps -flu oracle | grep SCAN => cd /proc/7835 (is the scan process id)

oracle@si01.an.com{LABDB1}/proc/7835> cat status

ipcs –m

------ Shared Memory Segments --------

key        shmid      owner      perms      bytes      nattch     status     

0x68d7e0dc 17727490   oracle    660        4096       0       

/usr/sbin/lsof –u oracle | grep 17727490

slabtop (sort by u: memory, 

strace -p 26175 -o pmon_strace.txt

Process 26175 attached - interrupt to quit

Process 26175 dettached

netstat –i:  Look for RX-OK, TX-OK fileds increasing on all interfaces.  RX-ERR and TX-ERR fields indicate if there is a fault and require investigation

netstat –-inet –a | more

Proto:  Protocol TCP / UDP

Recv-Q / Send-Q:  Receive queue and Send queue.  They should almost be zero.  If they’re increased on the UDP, it’s evident that your interconnect can’t sustain your desired workload.

Local Address              

Foreign Address            

State:  ESTABLISHED, LISTEN for TCP. 

ifconfig eth0:  Look for errors and drop packets

ss –l:  you can use the ss (socket statistic) as an alternative to netstat

tcpdump – D:  capture and display network packets 

1. eth0

2. eth1

tcpdump –i 2 | more