Transient outages, upstream faults, and service instability events.
Last reviewed: February 13, 2026|27 mapped errors|Page 1 of 2
Availability errors indicate that service execution failed or upstream dependencies could not complete in time for the current request path.
Provider-specific error pages mapped to this category (27 total). Showing 1-15.
AWS AccountProblem (Account Problem) means there is an AWS account-level issue that prevents the operation from completing and AWS directs you to contact Support. In Amazon S3, this error returns HTTP 403.
AWS EC2AccessDeniedException means Lambda needs additional permissions to configure VPC settings for the function invoke path (HTTP 502).
AWS InsufficientCapacityException means the service currently lacks enough backend capacity to satisfy the request. In services such as AWS Network Firewall, this can return as HTTP 500 and is usually transient.
AWS InternalError (Internal Error) means an internal Amazon S3 server error occurred while processing the request. In Amazon S3, this error returns HTTP 500.
AWS ServiceException means the AWS Lambda service encountered an internal error while processing the invoke request (HTTP 500).
AWS ServiceUnavailable (Service Unavailable) means Amazon S3 is temporarily unable to handle the request. In Amazon S3, this error returns HTTP 503.
AWS SubnetIPAddressLimitReachedException means Lambda could not set up VPC access because one or more configured subnets had no available IP addresses (HTTP 502).
AWS Unavailable (Unavailable) means the service is temporarily unavailable. In AWS APIs, this error returns HTTP 503.
Azure returns `ImageNotFound` when the VM image reference cannot be resolved for the selected region, subscription, or source registry.
ARM returns `NotFound` when a deployment operation references a resource that is unavailable at evaluation time.
Azure Table service returns `TableNotFound` when the target table name does not exist in the addressed storage account endpoint.
Azure returns `VMOSDiskNotFound` when a VM deployment references an OS managed disk identifier that cannot be resolved in the active scope.
GCP INTERNAL means the service detected an internal invariant failure while handling the request.
GCP UNAVAILABLE means the service is temporarily unable to handle the request; safe clients should retry with backoff.
GCP UNIMPLEMENTED means the called RPC, method, or capability is not implemented or not supported on the selected service endpoint.
Compare Guide
Use 429 for caller-specific throttling and 503 for service-wide outages, so retry behavior, escalation paths, and incident ownership stay correct.
Compare Guide
Debug 500 vs 502 faster: use 500 for origin failures and 502 for invalid upstream responses at gateways, then route incidents to the right team.
Compare Guide
Fix upstream errors faster: use 502 when a gateway gets an invalid upstream response, and 504 when the upstream service exceeds your timeout budget.
Playbook
Use this playbook to separate invalid upstream responses from upstream wait expiration and deadline exhaustion, and apply timeout budgets, safe retries, and circuit-breaker controls safely.
Playbook
Use this playbook to separate origin-side 500 failures from temporary 503 dependency or capacity outages, then apply safe retry and escalation paths.
Playbook
Triage 500, gRPC UNKNOWN, and cloud InternalError fast: preserve correlation IDs, separate transient provider faults from app bugs, and apply safe retries.
500 indicates internal server failure, 502 indicates invalid upstream response, 503 indicates temporary unavailability, and 504 indicates upstream timeout.
Retries worsen outages when they are unbounded or synchronized, because they increase load on already degraded dependencies.
Sustained normalization of error rate and tail latency under representative traffic, not only short success bursts.