Odd API error

[Chris]

Anyone else seen this?

 

I caught an exception:

 

[policyStatus] => DENY
[httpCode] => 503
[policyName] => loadProtection
[policyReason] => Load shedding threshold of 2.00 reached, rejecting for cool-off!

 

What does this mean, and how do I prevent it? Does it indicate a malicious user? (I know the user ID of the person that triggered this, and what page they were on). There wasn't an abnormal amount of activity at that time.

 

[Chris]

 

And another, this time:

 

 

[policyStatus] => DENY

[httpCode] => 503

[policyName] => loadProtection

[policyReason] => Load threshold of 4.00 exceeded!

 

I'm assuming they're to do with high load on a server, but the activity on the pages that calls this has been very light today. Less than 30 pageloads all day, and only 1 API call in the entire hour of this last error. The timings of these errors were 2011-05-19T15:12:06-07:00 and 2011-05-20T01:17:40-07:00

 

Has anyone else seen this? Does anyone know what causes it? Is it related to just my network's API activity, or all Ning API activity? 

 

 

 

[Tatu Saloranta]

This indicates that a load protection threshold was reached by one of the servers; and I think it is a global limit, not limited to your network activity. So it is not a direct indication that you would be doing too many calls. Further, since these limits are applied on per backend host basis, it does not actually indicate maximum rate you should be doing.

Limit is time-window based so once request rate drops enough, requests will go through again. So this is not a permanent block, and should be cleared up quite quickly (within seconds). This means that proper way to react to such errors is to add bit of delay before making a new request, similar to what should be done with other transient failures.

I hope this helps!

 

[Chris]

Thanks Tatu - that's very useful!