Ce que Repost a publié en avril 2026

Amazon Quick: How to disable per-action Allow/Deny prompts for connector Write actions (M365)?

[Urgent few days case hanging] Free tier case - 177722871100602

A user on Re:Post (AWS community forum) reported their Free Tier credits were revoked after a two-week inactivity period, despite updating account details. AWS support directed them to their existing …

How do I launch Appstream Desktop Application 2.0 after JumpCloud completes login to Appstream Web?

AWS AppStream 2.0 now supports redirecting streaming sessions from a web browser to the native AppStream client (WorkSpaces Applications client) using a 'amazonappstream:' prefixed URL after JumpCloud…

Secrets Store CSI Driver failing with Pod Identity on EKS Auto Mode

My support case stays in "Unassigned" status

A user reported a six-day delay in their AWS support case remaining unassigned, citing urgent email access issues tied to their corporate domain. AWS support staff acknowledged the frustration and adv…

Not able to connect from OnPrem/Local machine to EC2 instance.

Orchestrating ECS Task with Step function state macheine

AWS community clarifies how to pass tags to ECS tasks when orchestrating with Step Functions state machines. The correct method uses the Parameters field in the Task state definition, including a Tags…

Amazon QuickSight Access Error

Alternative factors not working

A user on AWS’s Re:Post forum reported that alternative MFA factors (e.g., email/phone verification) are not functioning as expected when signing in as a root user, with the prompt reappearing repeate…

AWS DMS (DB2 LUW source and S3 as target): Full load data appears to be “replaced” by CDC even with Target table preparation mode = DO_NOTHING

AWS DMS users report that full load data in S3 targets appears replaced by CDC changes even when TargetTablePrepMode is set to DO_NOTHING. AWS clarifies this is expected behavior due to S3's file-base…

How do you get out of the SES sandbox to send registration emails with Cognito

A user on Re:Post AWS reported a support case remaining unassigned for six days due to urgent corporate email access issues. AWS support advised verifying the support plan, checking case severity, and

Inconsistent k-NN result counts between primary and replica shards in Amazon OpenSearch Service. Are there any mitigations besides segment replication?

Amazon OpenSearch Service users report inconsistent k-NN vector search result counts between primary and replica shards due to non-deterministic HNSW graph construction. Segment replication is the onl…

S3 bucket on ME-South-1

Urgent billing dispute case unassigned for over a week, how can it be escalated?

A user on AWS’s Re:Post forum reported an unassigned billing dispute case (ID 177681062200855) lingering for 8 days with no response, risking overdue charges. Guidance suggests escalating via Support …

AWS IoT MQTT-over-WebSockets handshake returns 403 ForbiddenException for valid SigV4 with Admin IAM (us-east-2)

A user reported persistent 403 ForbiddenException errors during the WebSocket handshake for AWS IoT MQTT-over-WebSockets, despite valid SigV4 signatures and IAM permissions. REST-based IoT publishing …

Misleading error during 90-day password rotation: "old password incorrect" when new password fails policy

A user reported a misleading error message during AWS IAM password rotation where the system incorrectly blames an 'old password' failure when the new password violates policy requirements. The issue …

Best Practices for AWS WAF Rate Limiting with Mixed User Networks (Corporate + Public)

AWS re:Post published best practices for designing rate-limiting strategies in AWS WAF when users access applications from both corporate (shared IPs) and public networks. It recommends multi-layered …

Amplify Auth not handling authorization code from Entra redirect (manual token exchange required)

Users integrating Entra ID with AWS Cognito via Amplify Auth report that authorization codes from Entra-initiated SSO flows are not automatically handled, requiring manual token exchange. AWS confirms…

AWS DMS (DB2 LUW source and S3 as target): Full load data appears to be “replaced” by CDC even with Target table preparation mode = DO_NOTHING

Inconsistency between named device shadow document in cloud and in device

A user reported a persistent inconsistency between named device shadow documents in AWS IoT Greengrass, where updating a shadow value from 'false' to 'true' in the cloud fails to propagate to the devi

A user reported an outage in AWS ME-South-1 preventing S3 bucket access, with AWS re:Post recommending DataSync, S3 Replication, or CLI for data migration to other regions. The response notes versioni

Urgent billing dispute case unassigned for over a week, how can it be escalated?

AWS IoT MQTT-over-WebSockets handshake returns 403 ForbiddenException for valid SigV4 with Admin IAM (us-east-2)

Misleading error during 90-day password rotation: "old password incorrect" when new password fails policy

Best Practices for AWS WAF Rate Limiting with Mixed User Networks (Corporate + Public)

A user reported that Amplify Auth does not automatically process authorization codes from Entra ID redirects, requiring manual token exchange via Axios calls. AWS support clarified that Amplify is des

AWS Backup Restore Testing for EC2 always fails with "Restore metadata is invalid." due to NetworkInterfaces field

AWS Backup Restore Testing for EC2 consistently fails with 'Restore metadata is invalid' due to the NetworkInterfaces field in recovery point metadata. The field contains instance-specific values that…

AWS Step function state machine with ECS TASK

Users report inconsistent k-NN vector search result counts between primary and replica shards in Amazon OpenSearch Service due to non-deterministic HNSW graph construction. Segment replication is the

AWS Backup Restore Testing for EC2 resources consistently fails with 'Restore metadata is invalid' due to the NetworkInterfaces field in recovery point metadata. The field contains instance-specific v

Amazon Connect – Does queue delay on a higher priority queue block lower priority queues?

A user on AWS re:Post clarified that Amazon Connect’s queue delay mechanism does not block lower priority queues from being routed, even when higher-priority contacts are in delay. Contacts must first…

Assistance with AWS Lambda Node.js 20.x end-of-life for AWS-supported static website stack

A user reported receiving AWS Health Events about the Node.js 20.x runtime end-of-life for a Lambda function deployed via the AWS-supported static website stack (CDK-based). The issue requires updatin…

RestoreMetadataOverrides of my Cloudformation Template don't work

A user reported an issue with AWS Backup's RestoreMetadataOverrides in a CloudFormation template, causing restore testing to fail with an 'invalid metadata' error. Community responses suggested fixing…

cross account s3 access

s3 cross account access

OpenSearch service software update stuck in "Applying changes" for more than 6 months

A customer reports an OpenSearch Service domain stuck in 'UpdatingServiceSoftware' for months after a mandatory service-initiated update. The blue/green deployment halted at stage 4, leaving old and n…

Users integrating Entra ID with AWS Cognito via Amplify Auth report that authorization codes from Entra ID redirects are not automatically handled, requiring manual token exchange. AWS confirms Amplif

A user asked whether a higher-priority queue with a delay blocks lower-priority queues in Amazon Connect. Official documentation confirms that contacts in delay periods are not eligible for routing an

RestoreMetadataOverrides of my Cloudformation Template don't work

A user reported receiving AWS Health Events about the Node.js 20.x runtime end-of-life for a Lambda function deployed via the AWS-supported static website stack. The issue stems from a deprecated runt

A user reported that a mandatory OpenSearch Service software update has been stuck in the 'Applying changes' state for over six months, with the blue/green deployment halting at stage 4. The cluster r

A user reported receiving AWS Health Events about the Node.js 20.x runtime end-of-life for a Lambda function deployed via the AWS-supported static website stack sample. The issue requires updating the

A customer’s AWS OpenSearch Service domain has been stuck in an 'UpdatingServiceSoftware' state for over six months during a blue/green deployment. The deployment froze at stage 4, leaving both old an

AWS WAF: Safe Handling of Legit User IPs Without Blind Allowlisting

ECS express gateway deployment

SSM Distributor installing amazon-guardduty-agent 1.15.0 with non-idempotent postinst, breaks apt?

A non-idempotent postinst script in AmazonGuardDuty-RuntimeMonitoringSsmPlugin 1.15.0 causes apt to fail by leaving dpkg in a broken state. The script modifies systemd service files but fails on subse…

A user reported that Amazon GuardDuty agent version 1.15.0, distributed via SSM, includes a non-idempotent postinst script that modifies systemd service files. The script fails on subsequent runs if p

Urgent: Used AWS for learning purposes but recieved unexpected charges on my account and already opened a case ID but no response.

Request to Reinstate Promotional Credits

AWS DMS Serverless full load OOM (full load + CDC):best practices to reduce failures with cost constraints

Denied AWS Certification Exam Due to Name Mismatch – Test Center Raised Case – Request for Free Reschedule

CodeCommit shows "Complete your account setup" even though account is ACTIVE and fully verified

Amplify Custom Domain Stuck in AWAITING_APP_CNAME — SSL Certificate Verification Failing Despite Correct DNS

Transform and Marketplace - How do they relate, and who controls the available tools?

Amazon Connect AI Agent Designer not showing associated AgentCore Gateway MCP tool namespace

Users report that custom MCP tools from Bedrock AgentCore Gateway, such as lookup_policy, are not appearing in Amazon Connect AI Agent Designer despite successful integration setup. The issue stems fr…

Lambda CloudWatch log group not created when used existing role of other lambda

Users discovered that reusing an AWS Lambda execution role across multiple functions can prevent automatic CloudWatch log group creation for new functions. Logs may incorrectly route to an existing lo…

A user on AWS re:Post describes frequent out-of-memory (OOM) errors during AWS DMS Serverless full load migrations despite low DCU settings. Best practices recommended include splitting large migratio

A user reported that custom MCP tools from a Bedrock AgentCore Gateway Lambda target were not appearing in Amazon Connect AI Agent Designer despite successful integration association. The root cause w

CodeArtifact is unable to host @mui/icons-material 9.0.0

AWS re:Post users report CodeArtifact cannot host @mui/icons-material 9.0.0 because its package.json exceeds CodeArtifact’s 1MB metadata limit for npm publishes. The 1.37MB package.json triggers a har…

AWS re:Post users report CodeArtifact cannot host @mui/icons-material 9.0.0 because its package.json exceeds CodeArtifact's 1MB metadata size limit for npm publishing. The 1.37MB package.json triggers

Install Certbot on Amazon Linux 2023 (AL2023)

Best Practices for Aurora MySQL Advanced Auditing with Amazon RDS Proxy

AWS advises replacing fragile init_connect stored procedures with Aurora MySQL Advanced Auditing for compliance logging when using RDS Proxy. The new engine-level auditing eliminates dependency on rds…

Can I decrease the EBS volume size on my Amazon OpenSearch Service domain?

Amazon OpenSearch Service now permits reducing EBS volume size through a blue/green deployment, replacing the cluster with a smaller-volume version without downtime. This bypasses the standard EBS con…

EKS kubernetes service and inter-AZ traffic

The article clarifies how Kubernetes service, Endpoints, and EndpointSlice objects work in EKS, particularly for the default 'kubernetes' service. It reveals that these objects are managed by the upst…

Accelerating Incident Detection and Response onboarding with the Incident Detection and Response CLI

AWS introduced the Incident Detection and Response CLI (awsidr), a Python-based tool that automates workload registration, alarm creation, and APM integration for Incident Detection and Response. The …

Optimizing AWS DMS Replication Task Performance: Best Practices & Sample Configurations

AWS published a comprehensive guide to optimize AWS Database Migration Service (DMS) replication task performance, addressing common issues like slow full loads, CDC latency, and replication lag. The …

AWS Customer Guide: Leveraging global child safety partnerships for CSEA prevention

How do I automatically trigger AWS DevOps Agent investigations from CloudWatch Alarms and EventBridge rules for ECS tasks?

AWS published a guide enabling automatic triggering of DevOps Agent investigations via CloudWatch Alarms and EventBridge rules for ECS tasks. The integration uses Lambda functions with HMAC-authentica…

Integrating PagerDuty with Incident Detection and Response to improve critical workload monitoring

AWS announced an integration between its Incident Detection and Response service and PagerDuty to unify incident management for critical AWS workloads. The setup uses EventBridge and Lambda to route P…

How to Use a Dedicated IAM Role with the EFS CSI Driver Node Component in Amazon EKS Using Pod Identity

AWS detailed how to configure a dedicated IAM role for the EFS CSI driver node component in Amazon EKS using Pod Identity, enabling least-privilege access to EFS file systems. The article covers nativ…

Users report that custom MCP tools from Bedrock AgentCore Gateway Lambda targets are not appearing in Amazon Connect AI Agent Designer despite successful integration setup. The root cause is a misconf

AWS re:Post users report CodeArtifact cannot host @mui/icons-material 9.0.0 because its package.json exceeds CodeArtifact's 1MB metadata size limit for npm publishing. The issue stems from MUI's large

Applied for Toll-Free 15 business days ago still with no decision.

Ping status: Offline, Session Manager connection status: DisConnected for EC2 session manager connection

Users report EC2 Session Manager connection failures with errors indicating SSM Agent cannot acquire credentials or connect to AWS endpoints. Root causes include IMDSv2 hop limit restrictions, missing…

vSphere FT support on AWS EVS

TPM-WMI Event ID 1801 error on EC2 instance

A user reported a TPM-WMI Event ID 1801 error on Windows EC2 instances after March 2024 updates, related to Secure Boot DBX revocation list certificates not being applied to firmware. AWS support conf…

AWS re:Post users report CodeArtifact cannot host @mui/icons-material 9.0.0 because its package.json exceeds a 1MB metadata size limit for npm publishing. The 2GB total request limit cannot be raised,

Users report SSM Agent failures on AL2023 EC2 instances, showing 'Ping status: Offline' and 'Session Manager connection status: Disconnected' errors. Root causes include IMDSv2 hop limit restrictions,

A user reported a TPM-WMI Event ID 1801 error on multiple EC2 Windows instances following March 2024 Windows updates, related to Secure Boot DBX revocation list updates. AWS support clarified this is

Ubuntu Pro licensing for Lightsail instances — what's the supported path?

Can't receive inbound SMS via End User Messaging

Users on AWS re:Post report persistent failures receiving inbound SMS replies despite correct SNS topic and IAM configurations. Troubleshooting guidance highlights missing Pinpoint service principal i…

AWS CodeArtifact cannot host @mui/icons-material 9.0.0 because its package.json exceeds the 1MB metadata size limit for npm publishing. The 1.37MB package.json file triggers a hard constraint, despite

Users report Session Manager connection failures for AL2023 EC2 instances due to SSM Agent credential acquisition errors and network connectivity issues. Common causes include IMDSv2 hop limit restric

A user reported a TPM-WMI Event ID 1801 error on Windows EC2 instances following March 2024 Windows updates, related to Secure Boot DBX revocation list updates. AWS support confirmed it is a cosmetic

AWS re:Post clarifies that Lightsail instances cannot use Ubuntu Pro licensing, including for ESM Apps like ImageMagick patching. Users must migrate to EC2 for such needs, as Lightsail only supports i

Can't receive inbound SMS via End User Messaging

AWS support wait time to provision Bedrock quota

A user on AWS re:Post reports that Ubuntu Pro cannot be attached to Lightsail instances due to licensing restrictions, as Lightsail only supports its own bundles. The AWS response confirms this limita

A user reported failing to receive inbound SMS replies via AWS End User Messaging despite correct SNS topic and IAM configurations. AWS support suggested verifying sender country, topic policies, and

createRequestOriginGroup with originOverrides.hostHeader always causes 502 "invalid value for origin rewrite" regardless of value

A user reports that using originOverrides.hostHeader in CloudFront Functions' createRequestOriginGroup() with an Origin Group (S3 primary, ALB secondary) always triggers a 502 error, regardless of the…

"Curl error (27): Out of memory " after I update "dnf upgrade --releasever=2023.10.20260325"

Users on Amazon Linux 2023 EC2 instances encountered a 'curl error 27: out of memory' after running 'dnf upgrade' with a release version update. The issue stemmed from SSL/TLS configuration problems t…

301 redirect in lightsail

Users report a TPM-WMI Event ID 1801 error on EC2 Windows instances after applying updates, indicating unapplied Secure Boot certificates. The issue affects multiple instance types and requires manual

'Invalid principal in policy' error

AWS account suspended - Support not able to help

A user on AWS re:Post describes being unable to resolve an account suspension due to an expired credit card, citing broken support workflows, redirect loops, and unhelpful agents. They eventually foun…

A user reported that using originOverrides.hostHeader in CloudFront Functions' createRequestOriginGroup() with an Origin Group (S3 primary, ALB secondary) triggers a 502 error regardless of the hostHe

Users on Amazon Linux 2023 EC2 instances encountered a 'curl error (27): out of memory' after running 'dnf upgrade' with a releasever update. The error stemmed from SSL/TLS configuration conflicts, sp

A user on AWS re:Post describes being unable to resolve an account suspension due to an expired credit card, citing broken support workflows, redirect loops, and unresponsive agents. The issue was tem

I noticed a new rule in my WAF configuration, but I cannot identify the cause of its creation

A user on AWS re:Post discovered an automatically generated WAF rule named 'CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule' in their CloudFront distribution's Web ACL. AWS confirmed this occurs …

Formatting on comments on pull requests: CloudFormation Git Sync (Bitbucket)

A user reported that CloudFormation’s Git Sync feature splits change summaries into multiple Bitbucket comments, breaking Markdown table formatting in subsequent comments due to Bitbucket’s character …

A user reported failing to receive inbound SMS replies via AWS End User Messaging despite correct SNS topic and subscription setup. AWS support suggested verifying IAM roles, SNS topic policies, and e

A user reported that using originOverrides.hostHeader in CloudFront Functions' createRequestOriginGroup() with an Origin Group (S3 primary, ALB secondary) always triggers a 502 error, regardless of th

"Curl error (27): Out of memory " after I update "dnf upgrade --releasever=2023.10.20260325"

A user on AWS re:Post describes being unable to resolve an account suspension due to an expired credit card, citing broken support workflows where agents repeatedly redirect them to the console despit

CloudFront automatically generates a WAF rule named 'CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule' when geographic restrictions are enabled via the CloudFront console, even without manual WAF

Formatting on comments on pull requests: CloudFormation Git Sync (Bitbucket)

AWS Support Case for Billing Unassigned for more than 1 month

A user reported a billing support case opened on AWS re:Post in March 2026 remained unassigned for 15 days despite multiple follow-ups. Similar cases were resolved elsewhere, but this user's cases sta…

AWS re:Post published a comprehensive guide to optimize AWS Database Migration Service (DMS) replication task performance, addressing common issues like slow full loads, CDC latency, and task failures

AWS introduced a method to automatically trigger DevOps Agent investigations when CloudWatch Alarms enter ALARM state or EventBridge detects ECS task failures. The solution uses Lambda functions and H

AWS announced an integration between its Incident Detection and Response service and PagerDuty to unify incident management for critical workloads. The setup uses EventBridge and Lambda to route Pager

Getting Started with Kiro and MCP Servers: Connect Your AI IDE to Real-World Tools

AWS re:Post published a guide on integrating MCP servers with Kiro, an AI-powered IDE, to unlock advanced workflow automation. The article details installation, configuration, and verification steps f…

Meeting Cyber Essentials Patch Requirements for Amazon WorkSpaces Personal with AWS Systems Manager

AWS published a technical guide to automate patching for Amazon WorkSpaces in AUTO_STOP mode to meet the UK Cyber Essentials 14-day patching requirement. The solution uses AWS Systems Manager to orche…

Amazon Bedrock Inference Cost granularity based on IAM

Amazon Bedrock now captures IAM principal ARNs on every inference request, enabling per-caller cost attribution in CUR 2.0 billing exports. This eliminates manual workarounds like separate AWS account…

AWS DevOps Agent: A Technical Deep Dive into Autonomous Incident Response

AWS DevOps Agent is an automated incident response tool that reduces mean time to resolution (MTTR) from hours to minutes by performing root cause analysis during incidents. It operates through Agent …

AWS introduced the Incident Detection and Response CLI (awsidr), a Python-based tool that automates workload registration, alarm creation, and APM integration for its Incident Detection and Response s

A user on AWS re:Post describes a months-long struggle to resolve an AWS account suspension due to an expired credit card, citing unresponsive support, weeks-long case response times, and technical ba

A user discovered an undocumented CloudFront behavior where enabling geographic restrictions automatically creates a managed WAF rule named 'CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule' in as

A user reported a billing support case opened on AWS re:Post in March 2026 remained unassigned for 15 days, with follow-up attempts also unassigned. Testing with a root-user case showed the same issue

We’ve had incorrect VAT charged for ~2 years despite verified VAT. Two support cases, 2 weeks, no response. How do we get this assigned to AWS tax team?

An EU VAT-registered company reported being incorrectly charged local VAT on AWS invoices for two years despite verified VAT registration. Multiple support cases over two weeks received no human respo…

RME Monitron Sensor pairing issues

Users on AWS re:Post report inability to pair new RME Monitron sensors due to 'No NFC Scanner' errors, with some noting the issue stems from iOS app limitations. Commissioning sensors typically requir…

Users on AWS re:Post report a 'curl error (27): out of memory' after upgrading Amazon Linux 2023 instances using dnf. The error stems from SSL/TLS configuration issues tied to crypto-policies, not act

AWS account suspended - Support not able to help

CloudFront automatically generates a WAF rule named CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule when geographic restrictions are enabled via the CloudFront console, even without manual WAF da

AWS Support Case for Billing Unassigned for more than 1 month

An EU VAT-registered company claims AWS incorrectly charged local VAT for two years despite verified registration. Multiple support cases went unanswered for over two weeks, with no human response des

Users on AWS re:Post report inability to pair new RME Monitron sensors due to 'No NFC Scanner' errors, with some noting the issue stems from lack of NFC access in the iOS app. Commissioning sensors ty

Creating a Textract Dataset, are Queries only used?

A user on AWS re:Post describes a months-long struggle to resolve an AWS account suspension due to an expired credit card, citing unresponsive support, redirect loops, and login failures. They eventua

CloudFront automatically generates a WAF rule named 'CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule' when geographic restrictions are enabled in the CloudFront console, even without manual WAF d

A user reported that CloudFormation’s Git Sync feature splits pull request comments across multiple messages in Bitbucket, breaking formatting in the second comment due to Bitbucket’s character limit.

Users on AWS re:Post report inability to pair new RME Monitron sensors due to 'No NFC Scanner' errors, with some noting the issue stems from lack of NFC scanner access in iOS apps. Commissioning senso

AWS Support: Bedrock Kimi K2 / K2.5 Service Regression

AWS Bedrock’s Moonshot AI models (kimi-k2.5 and kimi-k2-thinking) in us-east-1 began returning all-padding output (!!!!) via the Converse API on 2026-04-23, exhausting token budgets without producing …

Amazon Bedrock’s Kimi K2.5 and Kimi K2 Thinking models in us-east-1 began returning only padding characters (!!!!) instead of valid responses via the Converse API on 2026-04-23, exhausting token budge

AWS Bedrock users report a high-severity regression in us-east-1 where Moonshot AI's kimi-k2.5 and kimi-k2-thinking models return only padding characters (!!!!) via the Converse API, exhausting token

Amazon Bedrock’s Kimi K2 and K2.5 models in us-east-1 began returning only padding characters (!!!...) via the Converse API on 2026-04-23, exhausting token budgets without producing usable output. The

AWS Bedrock’s Kimi K2 and K2.5 models in us-east-1 began returning only padding characters (!!!...) via the Converse API on 2026-04-23, exhausting token budgets without producing usable text. The issu

Users on AWS re:Post report being unable to pair new RME Monitron sensors due to 'No NFC Scanner' errors, with some noting the issue stems from iOS app limitations. Commissioning sensors typically req

An EU VAT-registered company claims AWS incorrectly charged local VAT for two years despite verified VAT registration. Multiple support cases over two weeks received no human response, and the issue r

Do Bedrock invoke commands support MiniMax models?

AWS confirmed that Bedrock's InvokeModel and InvokeModelWithResponseStream commands support MiniMax models (M2, M2.1, M2.5), including streaming via InvokeModelWithResponseStream. Additionally, the ne…

Amazon Bedrock "Operation not allowed" for ALL models despite full IAM permissions

Users report being unable to invoke any Amazon Bedrock models in us-east-1 due to a 'ValidationException: Operation not allowed' error, despite full IAM permissions. AWS support suggests the issue ste…

AWS CloudFront automatically generates a WAF rule named 'CreatedByCloudFrontForDistribution[ID]-Geo-BlockRule' when geographic restrictions are enabled via the CloudFront console, even without manual

An EU VAT-registered company claims AWS incorrectly charged local VAT for two years despite verified VAT registration. Multiple support cases over two weeks received no human response, and escalation

Users on AWS re:Post report being unable to pair new RME Monitron sensors due to 'No NFC Scanner' errors, with some noting the issue stems from limited iOS app functionality. Commissioning sensors typ

A user on AWS re:Post asks whether Textract dataset creation is solely for training queries or also for form fields. An answer clarifies that datasets are for training adapters to improve query respon

A user on AWS re:Post asked whether Bedrock's Invoke commands support MiniMax models, noting a lack of explicit documentation. An answer confirmed MiniMax models (M2, M2.1, M2.5) are supported via Inv

Amazon Bedrock "Operation not allowed" for ALL models despite full IAM permissions