I've been trying to learn container security for our ECS tasks, so I spun up Aqua's trial. For my small dev cluster, it's fine. But I just connected it to our main staging environment (~200 services) and... wow.
The console takes forever to load the vulnerability dashboard. Clicking into a specific image scan feels like it's hanging. Is this normal? My internet is fine, and other cloud consoles (AWS) are snappy.
I'm wondering if my Terraform setup is wrong somehow. I'm using the official Helm chart to deploy the console. Could it be a resource issue? We're on the default values.
You're definitely not the only one. The default resource requests in the Aqua Helm chart are often insufficient for any real deployment. At 200 services, you're hitting the point where the console's data aggregation becomes a bottleneck.
Check the memory usage on your console pods. I'd wager they're constantly hitting the limit, causing OOM kills and restarts during dashboard loads. Bump up the memory requests and limits, and consider scaling the console replicas if your license allows.
The latency usually comes from the console backend querying the Postgres database. For the vulnerability dashboard with that many images, it's often an unoptimized query with no useful indexes on large tables. You might need to tune the DB separately.
sub-10ms or bust
That's a really interesting point about the Terraform setup possibly being a factor. I've seen similar slow dashboard issues in other enterprise systems when the initial deployment spec doesn't account for data volume growth. Since you're on the default Helm values, it almost certainly is a resource issue, but I'm curious if the problem compounds based on *when* you connected the staging environment.
Did you deploy the console fresh and then immediately connect the 200-service environment? Or was it running with the dev cluster for a while first? I'm wondering if the console's internal database accumulates historical data or metadata in a way that bogs down performance over time, not just from a single large ingestion. A fresh install facing a huge data load might struggle differently than one that's been running.