Describe the bug
apify push fails consistently with a 503 Backend.max_conn reached error from a Varnish edge node, rather than reaching the application backend. This has persisted for 2+ hours and is fully reproducible.
Error output
Error: Status: 503
<html>
<head><title>503 Backend.max_conn reached</title></head>
<body>
<h1>Error 503 Backend.max_conn reached</h1>
<p>Backend.max_conn reached</p>
<h3>Error 54113</h3>
<p>Details: cache-fjr990024-FJR ...</p>
<hr>
<p>Varnish cache server</p>
</body>
</html>
To Reproduce
apify logincd <any actor project>apify push --force- Fails immediately with the above 503, every time
What I've tried (all failed identically)
- Retried 25+ times over 2+ hours, at increasing intervals
- Updated apify-cli from
1.2.1 -> 1.10.0 and retried
- Tried both
apify push and apify push --force
- Confirmed general API access works fine (
GET /v2/users/me, GET /v2/acts succeed normally) - only the build/source-upload path is affected
- status.apify.com shows "All Systems Operational," no incident reported
Apify support's response (via email, two separate replies) confirmed this is "an internal server-side issue on the API side" / "a temporary backend capacity or infrastructure issue" - not the CLI or account configuration - and noted the client's built-in retry logic (up to 8 retries, exponential backoff from 500ms) was already exhausted on every single attempt.
Environment
- apify-cli: 1.10.0 (also reproduced on 1.2.1)
- OS: Windows 11
- Node: 22.15.0
Always failing from the same Varnish node (cache-fjr990024-FJR), which points to an isolated backend/edge capacity issue rather than a CLI bug per se - filing here since the CLI is the user-facing symptom and other users may hit the same thing.
Describe the bug
apify pushfails consistently with a503 Backend.max_conn reachederror from a Varnish edge node, rather than reaching the application backend. This has persisted for 2+ hours and is fully reproducible.Error output
To Reproduce
apify logincd <any actor project>apify push --forceWhat I've tried (all failed identically)
1.2.1->1.10.0and retriedapify pushandapify push --forceGET /v2/users/me,GET /v2/actssucceed normally) - only the build/source-upload path is affectedApify support's response (via email, two separate replies) confirmed this is "an internal server-side issue on the API side" / "a temporary backend capacity or infrastructure issue" - not the CLI or account configuration - and noted the client's built-in retry logic (up to 8 retries, exponential backoff from 500ms) was already exhausted on every single attempt.
Environment
Always failing from the same Varnish node (
cache-fjr990024-FJR), which points to an isolated backend/edge capacity issue rather than a CLI bug per se - filing here since the CLI is the user-facing symptom and other users may hit the same thing.