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
I'd like to get support of multiple upstream hosts in single proxy + definition of balancing strategy (at least round-robin and/or least/most-connected).
As of now (v2.1.4) there is no way how to specify multiple upstreams for single proxy at least in toxiproxy-cli and python toxiproxy client.
Why (use-case)
My use-case uses toxiproxy to test the application's behaviour when introduced a toxics between the application and any of depending applications (all in kubernetes).
The way we introduce toxics w/o necessity of the application a change is to:
run toxiproxy and load all proxies to the dependent applications
force the application (app under test) to route via launched toxiproxy (via local linux /etc/hosts resolver)
Given the strategy it is clear that toxiproxy proxy binds same port as dependent app has/binds (no port translation in place).
Here comes the trouble, there is the application which communicates with HA storage consisting of multiple hosts binding the same port (redis sentinel), for such case I'd like to create toxiproxy proxy with multiple upstreams to block none / all upstreams.
The current workaround I have is to use multiple toxiproxy instances, but it comes with additional orchestration cost.
Any suggestions how to deal with the scenario are welcome.
The text was updated successfully, but these errors were encountered:
What (proposal)
I'd like to get support of multiple upstream hosts in single proxy + definition of balancing strategy (at least round-robin and/or least/most-connected).
As of now (
v2.1.4
) there is no way how to specify multiple upstreams for single proxy at least in toxiproxy-cli and python toxiproxy client.Why (use-case)
My use-case uses toxiproxy to test the application's behaviour when introduced a toxics between the application and any of depending applications (all in kubernetes).
The way we introduce toxics w/o necessity of the application a change is to:
/etc/hosts
resolver)Given the strategy it is clear that toxiproxy proxy binds same port as dependent app has/binds (no port translation in place).
Here comes the trouble, there is the application which communicates with HA storage consisting of multiple hosts binding the same port (redis sentinel), for such case I'd like to create toxiproxy proxy with multiple upstreams to block none / all upstreams.
The current workaround I have is to use multiple toxiproxy instances, but it comes with additional orchestration cost.
Any suggestions how to deal with the scenario are welcome.
The text was updated successfully, but these errors were encountered: