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
AFAIK the current the pool size configuration means `Max event processing threads` and this is per instance, and when Hasura runs horizontally this value could get multiplied.
What can go wrong?
Suppose If I have a event stream that goes down for some time
and when it comes back It could have significant load on the Hasura event queue
I would expect Hasura to unload that without having more than 100 (or whatever the poolsize) request at the same time.
Now if I have Hasura running horizontally (may be even with an auto scale setup) this could load my event handler in unexpectedly high rate.
(If I have 3 hasura instances up, Hasura would have 300 simultaneous requests, which I may not expect my RHS system to handle)
The event handler that I have in the RHS may not be designed to handle that much load, this could cause errors
Describe the solution you'd like
There should be a way to configure cumulative total max http poolsize instead of the event pool size. (just like total_max_connections)
Describe alternatives you've considered
If I have control over RHS, probably I can load balance RHS system or event increase the resources, but considering the expectation is to have Hasura manage the queue, this may not be a good solution.
The text was updated successfully, but these errors were encountered:
rahulagarwal13
changed the title
Provide a configuration to configure tatal max event poolsize (cumulative count over horizontal scaling)
Provide a configuration to configure total max event poolsize (cumulative count over horizontal scaling)
Dec 29, 2022
Is your proposal related to a problem?
AFAIK the current the pool size configuration means `Max event processing threads` and this is per instance, and when Hasura runs horizontally this value could get multiplied.
What can go wrong?
Describe the solution you'd like
There should be a way to configure cumulative total max http poolsize instead of the event pool size. (just like
total_max_connections
)Describe alternatives you've considered
If I have control over RHS, probably I can load balance RHS system or event increase the resources, but considering the expectation is to have Hasura manage the queue, this may not be a good solution.
The text was updated successfully, but these errors were encountered: