System-wide constraints on retries within distributed applications

A computer-implemented method, comprising:

• determining that a workflow comprising a plurality of inter-resource requests of a distributed application is to be initiated, wherein individual ones of the inter-resource requests of the plurality of inter-resource requests are to be directed from one resource of a plurality of resources to another resource of the plurality of resources, and wherein the plurality of resources includes a first resource, a second resource, and a third resource;
• in response to detecting, at the first resource during execution of the workflow, that (a) a first triggering condition for retrying a first inter-resource request directed to the second resource from the first resource has been satisfied and (b) a set of retry configuration settings of the workflow indicates that the first resource is permitted to retry the first inter-resource request, wherein the set of retry configuration settings is modifiable by at least some resources of the plurality of resources, wherein the set of retry configuration settings applies globally to the workflow as a unit:
• causing, by the first resource, the set of retry configuration settings to be modified to indicate a retry of the first inter-resource request;
• and
• retrying, by the first resource, the first inter-resource request;
• and
• in response to detecting, at the second resource during execution of the workflow after the retry configuration settings have been modified by the first resource, that (a) a second triggering condition for retrying a second inter-resource request directed to the third resource from the second resource has been satisfied and (b) the set of retry configuration settings indicates that the second resource is not permitted to retry the second inter-resource request, causing an indication that the workflow has failed to be generated, without retrying the second inter-resource request by the second resource.

One or more non-transitory computer-accessible storage media storing program instructions that when executed on or across one or more processors:

• determine that a workflow comprising a plurality of inter-resource requests of a distributed application is to be initiated, wherein individual ones of the inter-resource requests of the plurality of inter-resource requests are to be directed from one resource of a plurality of resources to another resource of the plurality of resources, and wherein the plurality of resources includes a first resource, a second resource, and a third resource;
• in response to detecting, at the first resource during execution of the workflow, that (a) a first triggering condition for retrying a first inter-resource request directed to the second resource from the first resource has been satisfied and (b) a workflow-level retry budget of the workflow indicates that the first resource is permitted to retry the first inter-resource request, wherein the work-level retry budget applies globally to the workflow, wherein the workflow-level retry budget is modifiable by at least some resources of the plurality of resources:
• retry, by the first resource, the first inter-resource request;
• and
• cause, by the first resource, the workflow-level retry budget to be modified to indicate that the first inter-resource request has been retried;
• and
• in response to detecting, at the second resource during execution of the workflow after the workflow-level retry budget has been modified by the first resource, that (a) a second triggering condition for retrying a second inter-resource request directed to the third resource from the second resource has been satisfied and (b) the workflow-level retry budget, which was modified at least by the first resource, indicates that the second resource is not permitted to retry the second inter-resource request, cause an indication that the workflow has failed to be generated, without retrying the second inter-resource request by the second resource.