As part of cost optimizaiton initatives, we were trying to set some cpu/memory limits for our temporal pods in dev environment.
we did not have any resource configurations previously and now introduced the following ,limits
typically .1 cpu and 128Mi to 1Gi
| temporal-frontend | CPU | Memory | Replica Count | ||
|---|---|---|---|---|---|
| Min | Max | Min | Max | ||
| temporal-admintools | 0.1 | 0.1 | 128 Mi | 128 Mi | 1 |
| temporal-matching | 0.1 | 0.1 | 128 Mi | 1 Gi | 1 |
| temporal-history | 0.1 | 0.1 | 128 Mi | 1 Gi | 2 |
| temporal-worker | 0.1 | 0.1 | 128 Mi | 1 Gi | 2 |
| temporal-frontend | 0.1 | 0.1 | 128 Mi | 1 Gi | 2 |
After the resource configs were introduced we started seeing
"temporal server pods I saw the error “no usable DB connections found” in the temoral-server pods.
any hints on how to size the cpu/memory for the pods,
we may have about 10000 to 40000 workflows running max 2 work spaces.
the workflows are long living.
some of the workflows may are nested (say 3 levels of children workflows),
most worklows may have upto 22 activities.