An Azure native disaster recovery service. Previously known as Microsoft Azure Hyper-V Recovery Manager.
A private cache endpoint is supported for Azure-to-Azure replication, so the next decisive check is an authenticated AzCopy upload from the source VM to the cache account; this distinguishes a working TCP handshake from the authorization, storage-policy, routing, or proxy failure represented by error 153017.
1. Reproduce the actual storage operation
Test-NetConnection proves only that TCP 443 opens. It does not prove that the Mobility service can authenticate and write a blob.
- From the source VM, use AzCopy to upload a test file to a temporary container in the cache account.
- Use the same Blob FQDN that resolves to the private endpoint.
- Record the HTTP result and request ID:
-
403points toward authorization, storage policy, or the request reaching an unintended endpoint. - Timeout/reset points toward routing, firewall, proxy, or NVA handling.
- A successful upload shifts the investigation to the ASR agent/service operation.
-
Microsoft specifically recommends AzCopy from the source VM for testing uploads to the cache account and identifying latency or NVA-path problems.^1^
Example using interactive Microsoft Entra authentication and a temporary container:
'ASR upload test' | Set-Content C:\Temp\asr-test.txt
azcopy login
azcopy copy `
'C:\Temp\asr-test.txt' `
'https://<cache-account>.blob.core.windows.net/<test-container>/asr-test.txt' `
--overwrite=true `
--log-level=DEBUG
This test validates the user’s identity, not the vault identity, but it confirms whether a complete authenticated Blob request can traverse the source VM’s actual network path. Preserve the AzCopy debug log and storage request ID.
2. Validate the supported cache configuration
Check Storage account → Overview/Configuration/Data protection:
- The account must be General-purpose v2 for a storage private endpoint.
- It must be in Central India, the same region as the protected VM.
- Blob soft delete must be disabled; soft delete is not supported for an ASR cache account.
- The cache container/account must still exist and be the resource selected in the replicated item.
- The Blob private endpoint connection must show Approved.
These are Azure-to-Azure cache requirements.^2^ A storage private endpoint is supported, but it must be for the cache/log account in the source network and have network line of sight from the source machine.^3^
Because DNS and TCP tests already succeed, do not repeat those checks. Instead, verify the path actually used:
- Private endpoint → Metrics → Bytes In/Bytes Out while reproducing the error.
- Check effective routes on the source VM NIC for the private endpoint IP.
- If a proxy is configured, bypass it for the exact cache FQDN. Private-endpoint traffic must follow the VNet route, not be sent to an outbound web proxy.
- Check whether an NVA or forced-tunnel route is intercepting replication traffic. Microsoft notes that an NVA can throttle or disrupt cache uploads.^4^
3. Authentication and shared-key access
Disabling shared-key access is supported. ASR can use a cache account with key-based authentication disabled when:
- Managed identity is enabled on the Recovery Services vault.
- That identity has read/write access to the cache account.
Therefore, shared-key access should not need to be re-enabled solely because of 153017.^5^
Since the managed identity and data-plane role are already configured, inspect Storage account → Activity log for failed role-assignment changes and enable Blob diagnostic logging to capture failed requests. A 403 with its storage error code is substantially more useful than 153017 alone.
4. Required outbound connectivity
The cache private endpoint is not the Mobility agent’s only dependency. The source VM also requires outbound HTTPS access to:
- The cache account Blob FQDN
-
login.microsoftonline.com -
*.hypervrecoverymanager.windowsazure.com -
*.servicebus.windows.net -
*.vault.azure.netfor ADE-enabled VMs -
*.automation.ext.azure.comwhen Mobility agent automatic upgrade is enabled
Authenticated outbound proxies are not supported, and IP-based filtering is not recommended; use the documented service tags or URL allow-list.^4^ Microsoft Entra ID itself does not support private endpoints, so it still requires permitted outbound access even when both the vault and cache use Private Link.^3^
5. Logs and escalation
The published svagents*.log locations apply to VMware/physical-server Mobility agents, not Azure-to-Azure VM replication, so they should not be treated as the authoritative log path for this case.
For Azure-to-Azure:
- Open Recovery Services vault → Site Recovery jobs.
- Select the failed operation.
- Use ExportJob → Export and retain the exported job details.^6^
- Configure Vault → Diagnostic settings → Add diagnostic setting, send all
AzureSiteRecovery*categories to Log Analytics, then reproduce the failure.^7^ - If AzCopy succeeds but ASR continues returning 153017, open an Azure support request and attach:
- Exported Site Recovery job
- Exact failure timestamp and correlation/request IDs
- AzCopy debug log
- Storage Blob diagnostic events
- Effective route table and proxy/NVA details
References
- Troubleshoot replication in Azure VM disaster recovery
- Support matrix for Azure VM disaster recovery between Azure regions
- Replicate machines with private endpoints in Azure Site Recovery
- About networking in Azure virtual machine disaster recovery
- Turn off key based access on cache account
- How to troubleshoot Azure Recovery Services
- Monitor Site Recovery with Azure Monitor Logs