You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If all replica instances are down. Current setup will NOT route those traffic to the only primary. which means readonly service are not available.
It nice to have an additional option to configure that in case of all replicas failure:
Leave primary be, so at least readwrite service will survival the avalanche case
Route readonly traffic to primary so the service still intact.
Proposal
Add an option haproxy_weight_fallback which controls this behaviro
When set to 0, it is same as before, primary will not take any read-only traffic
When set to a value larger than 0, It will be used as relative weight for primary instance in replica services group.
In normal situation, the primary will take a very little portion of read-only traffic. In case of all replica down. It will become the only available backend in replica service. And takeover all read-only traffics.
The text was updated successfully, but these errors were encountered:
Scenario
If all replica instances are down. Current setup will NOT route those traffic to the only primary. which means readonly service are not available.
It nice to have an additional option to configure that in case of all replicas failure:
Proposal
Add an option
haproxy_weight_fallback
which controls this behaviroIn normal situation, the primary will take a very little portion of read-only traffic. In case of all replica down. It will become the only available backend in replica service. And takeover all read-only traffics.
The text was updated successfully, but these errors were encountered: