|
Re:sea failback - delay issue (1 viewing) (1) Guest
Favoured: 36
|
|
|
TOPIC: Re:sea failback - delay issue
|
iulian (User)
Junior Boarder
Posts: 34
|
|
sea failback - delay issue 11 Months, 2 Weeks ago
|
|
|
Hi everybody,
I have a dual vio configuration and i noticed that if primary VIO is rebooted , there is a delay of approx. 30-40 sec for network failback. This is probably because of the spanning tree on the switches (i use etherchannel for the physical adapters which are part of sea). It seems to works as designed.
My question is if there is any workaround to DISABLE sea failback ?
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
Claus (Moderator)
Moderator
Posts: 29
|
|
Re:sea failback - delay issue 11 Months, 2 Weeks ago
|
|
|
Hello Iulian,
you can set the ha_mode to standby
chdev -dev entX -attr ha_mode=standby
which performs a failover to the other SEA. With this settings you can enforce the system to use the other SEA even if it has the lower trunc priority. To allow a fallback change the ha_mode back to auto.
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
iulian (User)
Junior Boarder
Posts: 34
|
|
Re:sea failback - delay issue 11 Months, 2 Weeks ago
|
|
|
Hi Claus,
I am aware about the manual failover/failback possibilities. What i want to know is if it is possible to disable failback when the primary is rebooted ? I have tried some manual odm tricks to change settings for SEA on phase 2, but it seems it doesn't work. The problem is that after the primary VIO is rebooted, it automatically failback once the SEA is again available, which is not what I want (because of the 30-40 sec delay).
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
iulian (User)
Junior Boarder
Posts: 34
|
|
Re:sea failback - delay issue 11 Months ago
|
|
|
Well, it seems that it is not possible to disable the SEA failback so far. There are many requests to IBM to introduce this option as a SEA attribute. The only possibility is to use a workaround (small script which is triggered in cfgmgr phase 2, which will change the ha_mode to standby).
Did anyone managed to configure SEA with link aggregation and to get minimum delay (zero ping packet loss) during failback (after primary VIO is rebooted) ?
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
jgan (User)
Fresh Boarder
Posts: 18
|
|
Re:sea failback - delay issue 11 Months ago
|
|
|
some switch brands allow a relaxed handling of spanning tree.
for cisco check out "portfast"
and "bpdu guard" settings.
or use NIB in the vio client. with multiple virtual switches (p6+) this even allows bridging tagged vlans today.
cheers
Joachim
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
iulian (User)
Junior Boarder
Posts: 34
|
|
Re:sea failback - delay issue 10 Months ago
|
|
|
The problem has been solved by using standard aggregation method instead of 8023.ad . I have tried all possibilities and settings on the adapter/switch and the minimum delay i could get when using 8023.ad aggregation was 5 sec.
|
|
|
|
|
|
|
The administrator has disabled public write access.
|
|
|
|
|
|
|
FORUM 
