Uunexpected-ivory
Hi everyone new to Apify and I’m stuck
Hi everyone new to Apify and I’m stuck on a concurrency issue.
I want to run 2+ runs of a Store Actor in parallel (same Actor ID, different inputs) from Python.
What happens: my code submits multiple runs “at once” (I’ve tried both threads and asyncio), but the second run appears to “stall” for minutes and only begins to send actor’s input after the first run finishes. During the stall it only shows container startup lines like:
ACTOR: Pulling container image...
ACTOR: Creating container.
ACTOR: Starting container.
Then later (after run #1 ends) it finally sends the actor (proxy/arguments/etc.).
Both runs show as "RUNNING" state at all times
Account / limits: plan shows 8 GB Total Actor RAM and 25 concurrent runs, and I’m running each at 256 MB; only 2 runs active during the test.
Question: could this be:
A per-user concurrency lock inside the Actor (e.g., the Actor prevents multiple runs per user / uses a global lock / shared state), or
An Apify platform behavior (queueing / worker allocation), even though I’m within RAM + concurrent run limits?
I want to run 2+ runs of a Store Actor in parallel (same Actor ID, different inputs) from Python.
What happens: my code submits multiple runs “at once” (I’ve tried both threads and asyncio), but the second run appears to “stall” for minutes and only begins to send actor’s input after the first run finishes. During the stall it only shows container startup lines like:
ACTOR: Pulling container image...
ACTOR: Creating container.
ACTOR: Starting container.
Then later (after run #1 ends) it finally sends the actor (proxy/arguments/etc.).
Both runs show as "RUNNING" state at all times
Account / limits: plan shows 8 GB Total Actor RAM and 25 concurrent runs, and I’m running each at 256 MB; only 2 runs active during the test.
Question: could this be:
A per-user concurrency lock inside the Actor (e.g., the Actor prevents multiple runs per user / uses a global lock / shared state), or
An Apify platform behavior (queueing / worker allocation), even though I’m within RAM + concurrent run limits?
