Hi again Steven,
Ok I think I understand the thing, I've tried several values of token
and got these results :
token: 5000 token_retransmits_before_loss_const: 20
consensus: 7500
==> ~10s
token: 10000 token_retransmits_before_loss_const: 20
consensus: 12000
==> ~12s
token: 15000 token_retransmits_before_loss_const: 20
consensus: 18000
==> ~36s
token: 20000 token_retransmits_before_loss_const: 20
consensus: 24000
==> ~45s
token: 25000 token_retransmits_before_loss_const: 20
consensus: 30000
==> ~55s
so I think that only token value (and dependant consensus = 1.2 token)
is only meaningful for
this heartbeat timeout before deciding to fence.
And that you suggest to increase the token_retransmits_before_loss_const
just because
there is more time (as we increase token) to give more tries for
retransmit ,
Am I right ?
Thanks
Regards.
Alain
_______________________________________________
Openais mailing list
Openais@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/openais
