I am a paying customer and every now and then the server starts replying with “HTTP/1.1 503 Service Unavailable: Back-end server is at capacity”.
This only happens for a short period of time (minutes) and then everything comes back to normal … any thoughts?
Sample curl request - not sure if it helps:
HTTP/1.1 503 Service Unavailable: Back-end server is at capacity
Content-Type: text/plain
Date: Sun, 19 Aug 2018 15:40:59 GMT
Server: Mashape/5.0.6
X-RateLimit-requests-Limit: 25000
X-RateLimit-requests-Remaining: 19426
加入讨论 - 在下面添加评论:
I was able to put a script together to reproduce. Could you keep it to 5 concurrent connections or less? I think that will solve your problem.
I don’t see any errors on the backend servers so I just turned on logging for the load balancer in front of them. I also bumped up the number of backend server nodes as well to help with the volume. I’ll keep an eye on the load balancer logs and see if I can spot anything. Can you attempt to reproduce?
Ok, so the new error I encountered is “HTTP/1.1 504 GATEWAY_TIMEOUT” just now. But … from my tests, I can deduct that it happens only during high load - eg: when if I am doing a lot of concurrent calls (50+ maybe). Could that be the case?
If yes then can you recommend an optimal number of concurrent calls?
Hi, thanks for your support! Will keep an eye on it and get back should it happen again!
Hi - I wasn’t able to reproduce but I noticed some errors in the httpd logs due to a bug. Fixed the bug and just deployed. Can you check if you are still able to reproduce the issue? Thanks -
Hi - it was fine for a while but now the issue seems to be back. Sample request:
curl --get --include ‘https://domaination.p.mashape.com/domains/somedomain.com’ -H ‘X-Mashape-Key: XXX’ -H 'Accept: application/json’
HTTP/1.1 503 Service Unavailable: Back-end server is at capacity
Content-Type: text/plain
Date: Mon, 27 Aug 2018 06:24:33 GMT
Server: Mashape/5.0.6
X-RateLimit-requests-Limit: 25000
X-RateLimit-requests-Remaining: 5192
Content-Length: 0
Connection: keep-alive
Please advise!
Hi Lucian - I believe this was a temporary issue due to servers having reached capacity. The backend is supported by an autoscaling aws cluster but sometimes during spikes capacity is reached while it scales up. I increased the number of servers in case this helps. Are you still having issues?