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
Some recent ClamAV benchmarking work demonstrates that AV scan performance degrades substantially as the number of concurrent scan requests increases. To minimize unacceptable long upload times as traffic increases in March, we should spin up redundant ClamAV instances so that we can split the load between them.
There won't be an actual load balancer mediating the traffic between the FAC and the ClamAV instances, so we'll need to implement some rudimentary load balancing inside the application. For our purposes, it should suffice to pick randomly from the set of available ClamAV instances rather than complicate the picture by trying round-robin.
We're done when
The AV scan step of the file upload pipeline is routed randomly to one of the configured ClamAV instances
Over time, the requests to the ClamAV instances distributed (approximately) uniformly
The FAC continues to work as expected with a single ClamAV instance
The text was updated successfully, but these errors were encountered:
Epic: #3358
Background
Some recent ClamAV benchmarking work demonstrates that AV scan performance degrades substantially as the number of concurrent scan requests increases. To minimize unacceptable long upload times as traffic increases in March, we should spin up redundant ClamAV instances so that we can split the load between them.
There won't be an actual load balancer mediating the traffic between the FAC and the ClamAV instances, so we'll need to implement some rudimentary load balancing inside the application. For our purposes, it should suffice to pick randomly from the set of available ClamAV instances rather than complicate the picture by trying round-robin.
We're done when
The text was updated successfully, but these errors were encountered: