Dengan HADR
-------------------
1) Put dummy IP on top of SEA interfaces on above VIOS 1 and VIOS 2:
ifconfig en21 0.0.0.0 up (only run on VIOS 1)
ifconfig en29 0.0.0.0 up (only run on VIOS 2)
2) Clear out snap data and collect new, initial data on all systems (VIOS 1, VIOS
2, LPAR 1 and LPAR 2):
snap -r
mkdir -p /tmp/ibmsupt/testcase/hadr/
date > /tmp/ibmsupt/testcase/hadr/date.before
ifconfig -a > /tmp/ibmsupt/testcase/hadr/ifconfig.out
no -a > /tmp/ibmsupt/testcase/hadr/no.out
netstat -s > /tmp/ibmsupt/testcase/hadr/netstat.s.before
netstat -v > /tmp/ibmsupt/testcase/hadr/netstat.v.before
netstat -in > /tmp/ibmsupt/testcase/hadr/netstat.in.before
netstat -rn > /tmp/ibmsupt/testcase/hadr/netstat.rn.before
netstat -ano > /tmp/ibmsupt/testcase/hadr/netstat.ano.before
arp -an > /tmp/ibmsupt/testcase/hadr/arp.before
errpt -a > /tmp/ibmsupt/testcase/hadr/errpt.before
3) Run tcpdump and lparstat on above mentioned VIOS:
VIOS 1:
tcpdump -i en21 -s 128 -C 200 -W 12 -w /tmp/ibmsupt/testcase/hadr/tcpdump_en21.cap
&
lparstat -h -t 5 > /tmp/ibmsupt/testcase/hadr/lparstat.out &
VIOS 2:
tcpdump -i en29 -s 128 -C 200 -W 12 -w /tmp/ibmsupt/testcase/hadr/tcpdump_en29.cap
&
lparstat -h -t 5 > /tmp/ibmsupt/testcase/hadr/lparstat.out &
NOTE: Keep track of the process IDs
4) Run tcpdump and lparstat on AIX LPAR 1 and AIX LPAR 2:
tcpdump -i en1 -s 128 -C 100 -W 12 -w /tmp/ibmsupt/testcase/hadr/tcpdump_en1.cap &
lparstat -h -t 5 > /tmp/ibmsupt/testcase/hadr/lparstat.out &
NOTE: Keep track of the process IDs
5) Re-create the slowness issue (whether it's with scp or batch job application ).
6) As soon as slowness issue is captured, stop packet captures on both VIOS and
both LPARs:
You can kill the tcpdump and lparstat via kill command (not "kill -9"):
kill <tcpdump_pid>
kill <lparstat_pid>
(Process IDs are from steps 3 and 4)
7) Collect new data on LPAR 1, LPAR 2, VIOS 1 and VIOS2:
date > /tmp/ibmsupt/testcase/hadr/date.after
netstat -s > /tmp/ibmsupt/testcase/hadr/netstat.s.after
netstat -v > /tmp/ibmsupt/testcase/hadr/netstat.v.after
netstat -in > /tmp/ibmsupt/testcase/hadr/netstat.in.after
netstat -rn > /tmp/ibmsupt/testcase/hadr/netstat.rn.after
netstat -ano > /tmp/ibmsupt/testcase/hadr/netstat.ano.after
arp -an > /tmp/ibmsupt/testcase/hadr/arp.after
errpt -a > /tmp/ibmsupt/testcase/hadr/errpt.after
8) Detach the SEA interface on VIOS 1 and VIOS 2:
ifconfig en21 down; ifconfig en21 detach (NOTE: replace enX with interfaces that
were put up on Step 1)
9) Tar the testcase folders and upload to IBM. Please provide the TCP port being
used for the application during the above run.
startsrc -s iptrace -a "-L 300000000 -b /tmp/TS017051501.$(hostname).iptrace.bin"
-Reproduce issue: initiate a scp transfer to from AIX to Linux to recreate the
disconnection
stopsrc -s iptrace The above syntax will rotate files once the file size reaches
300/tmpMB.
nohup /tmp/perf72/perfpmr.sh &
pax -xpax -vw perfdata | gzip -c > sloweod20240911.perfdata.dbsyariah1jkt.pax.gz
pax -xpax -vw perfdata | gzip -c > sloweod20240911.perfdata.dbsyariah1sby.pax.gz
