Testy IPMP/Solaris 10

IPMP – IP MultiPathing to technika HA ( High Avability ) pozwalajaca utrzymac lacznosc ze swiatem serwera z systemem Solaris mimo awarii jednego lub wiecej polaczen sieciowych. To tyle tytulem wprowadzenia :)

xeno – komputer podlaczony do switcha pojedyncza karta sieciowa
10.99.1.20 – Solaris10 z IPMP ( 2 karty sieciowe do tego samego switcha co xeno )

xeno:~# ping -i 0.5 10.99.1.20
PING 10.99.1.20 (10.99.1.20) 56(84) bytes of data.
64 bytes from 10.99.1.20: icmp_seq=1 ttl=255 time=0.318 ms
64 bytes from 10.99.1.20: icmp_seq=2 ttl=255 time=0.308 ms
64 bytes from 10.99.1.20: icmp_seq=3 ttl=255 time=0.293 ms
64 bytes from 10.99.1.20: icmp_seq=4 ttl=255 time=0.325 ms
64 bytes from 10.99.1.20: icmp_seq=5 ttl=255 time=0.312 ms
64 bytes from 10.99.1.20: icmp_seq=6 ttl=255 time=0.308 ms
64 bytes from 10.99.1.20: icmp_seq=7 ttl=255 time=0.342 ms
64 bytes from 10.99.1.20: icmp_seq=8 ttl=255 time=0.325 ms
# wyciagniecie kabelka...
64 bytes from 10.99.1.20: icmp_seq=19 ttl=255 time=0.267 ms
64 bytes from 10.99.1.20: icmp_seq=20 ttl=255 time=0.271 ms
64 bytes from 10.99.1.20: icmp_seq=21 ttl=255 time=0.338 ms
64 bytes from 10.99.1.20: icmp_seq=22 ttl=255 time=0.280 ms
64 bytes from 10.99.1.20: icmp_seq=23 ttl=255 time=0.254 ms

Czyli 19-8 pakietow zgubionych, co przeklada sie na (19-8)*0.5=5.5 sekund niedostepnosci maszyny przy ponizszym wpisie do /etc/default/mpathd:
# 2s
FAILURE_DETECTION_TIME=2000

A teraz kilka nieudokumentowanych flag dla in.mpathd:

  • -a “adopt” mode
  • -f force multicast
  • -d debug/foreground
  • -D debug flags
  • -l Turn off link state notification handling. ( debug )

Comments are closed.