Please help with command to check the busy % for a hard disk device like hdisk5
Best regards,
Vishal
Have a look at the manual page for df if you mean full, or sar for IO performance - probably the -d flag, but you haven't given us the OS, so it's a bit difficult to be sure.
Robin
u can also use topas
This is AIX .Below is topas information but it's not giving for each hdd .Please help
CPU User% Kern% Wait% Idle% Reads 1613 Rawin 0
Total 6.4 4.7 18.6 70.3 Writes 26 Ttyout 465
Forks 4 Igets 0
Network BPS I-Pkts O-Pkts B-In B-Out Execs 4 Namei 1743
Total 282K 270.5 181.0 208K 74.5K Runqueue 5.00 Dirblk 0
Waitqueue 0.0
Disk Busy% BPS TPS B-Read B-Writ MEMORY
Total 7.8 5.63M 692.5 5.31M 322K PAGING Real,MB 12288
Faults 4122 % Comp 73
FileSystem BPS TPS B-Read B-Writ Steals 0 % Noncomp 17
Total 7.16M 1.59K 6.86M 308K PgspIn 0 % Client 17
PgspOut 0
Name PID CPU% PgSp Owner PageIn 0 PAGING SPACE
osysmond 9175044 1.4 121M root PageOut 0 Size,MB 8192
oracle 43778058 1.2 12.0M oracle Sios 0 % Used 21
oraroota 9568384 0.9 32.4M root % Free 79
oracle 23199944 0.8 9.57M oracle NFS (calls/sec)
oracle 31129738 0.7 10.9M oracle SerV2 0 WPAR Activ 0
oracle 55771278 0.6 12.1M oracle CliV2 0 WPAR Total 0
oracle 24183012 0.6 10.9M oracle SerV3 0 Press: "h"-help
oracle 49479914 0.4 12.1M oracle CliV3 0 "q"-quit
oraagent 9830536 0.3 70.6M oracle
Best regards,
Vishal
You could try nmon
After it's started, press a lower case d
Still probably better with sar -d 5 | grep "hdisk5 " which you can trap to a file and shovel into something like Excel to draw graphs if that is useful to understand.
This will give you the IO rates every 5 seconds. Colume order is:-
time device %busy avque r+w/s Kbs/s avwait avserv
The time may not be displayed though. If you want to force a timestamp, you might need to do something like:-
echo "`date +%d/%m/%Y %T` `sar -d 5 2 | grep "hdisk5 " | tail -1` > logfile
Robin
when you are in the topas screen
type d
---------- Post updated at 02:58 PM ---------- Previous update was at 02:58 PM ----------
man topas :
---------- Post updated at 02:59 PM ---------- Previous update was at 02:58 PM ----------
Topas Monitor for host: delta EVENTS/QUEUES FILE/TTY
Thu Jan 9 14:59:48 2014 Interval: 2 Cswitch 233 Readch 4079
Syscall 117 Writech 3252
CPU User% Kern% Wait% Idle% Physc Entc Reads 29 Rawin 0
ALL 0.3 0.7 0.0 99.0 0.01 1.6 Writes 4 Ttyout 830
Forks 0 Igets 0
Network KBPS I-Pack O-Pack KB-In KB-Out Execs 0 Namei 11
Total 2.2 4.5 6.0 0.2 1.9 Runqueue 0.0 Dirblk 0
Waitqueue 0.0
Disk Busy% KBPS TPS KB-Read KB-Writ MEMORY
hdisk0 0.0 0.0 0.0 0.0 0.0 PAGING Real,MB 704
cd0 0.0 0.0 0.0 0.0 0.0 Faults 0 % Comp 95
cd1 0.0 0.0 0.0 0.0 0.0 Steals 0 % Noncomp 1
PgspIn 0 % Client 1
FileSystem KBPS TPS KB-Read KB-Writ PgspOut 0
Total 4.0 27.5 4.0 0.0 PageIn 0 PAGING SPACE
PageOut 0 Size,MB 512
Name PID CPU% PgSp Owner Sios 0 % Used 6
topas 516342 0.4 1.9 root % Free 94
getty 311500 0.1 0.6 root NFS (calls/sec)
nfsd 286882 0.1 0.2 root SerV2 0 WPAR Activ 0
sshd 467026 0.0 1.1 root CliV2 0 WPAR Total 0
gil 53274 0.0 0.1 root SerV3 0 Press: "h"-help
random 180340 0.0 0.1 root CliV3 0 "q"-quit
java 278724 0.0 51.1 root
java 213174 0.0 36.6 pconsole
rpc.lock 225282 0.0 0.2 root
pilegc 36882 0.0 0.1 root
dirsnmpd 172186 0.0 7.1 root
netm 49176 0.0 0.1 root
syncd 98440 0.0 0.6 root
lrud 16392 0.0 0.1 root
topasrec 307440 0.0 0.8 root
IBM.CSMA 360636 0.0 2.2 root
sendmail 200860 0.0 1.1 root
init 1 0.0 0.7 root
cimserve 315578 0.0 23.6 root
xmgc 40980 0.0 0.1 root
Note that the interpretation of metrics like disk utilization's busy and util should be different for modern architectures that use external storage (SAN, NAS etc.).
This post talks about Linux, but I suppose that AIX's not much different in this regard.
1 Like
For topas, you can also just feed it the -D switch to only show the hdisk info:
topas -D