You will find the following error in the job log:
[timestamp] Error Failed to expand object added to the job. Object: "Object, id: 7de79ed1-19fa-410e-af90-33bc579cfd07, name: VMname, type: VM, vi type: VirtualMachine, host name: esxhost".
[timestamp] Error Failed to login to "VC" by SOAP, port 443, user "user", proxy srv: port:0 (Veeam.Backup.Common.CommonException)
[timestamp] Error The request failed with HTTP status 503: Service Unavailable. (System.Net.WebException)
[timestamp] Error at System.Web.Services.Protocols.SoapHttpClientProtocol.ReadResponse(SoapClientMessage message, WebResponse response, Stream responseStream, Boolean asyncCall)
The Backup & Replication job starts to fail suddenly with "failed to login to "<vcenter server>" by SOAP, port 443, user "<SOAP account>" each time, however it is possible to login to that vCenter using the vSphere Client. You are able to expand the hierarchy of it in the Veeam Console. If you reboot the vCenter server, the job will succeed, but after a period of time, the job will fail with the same error again.
The vCenter database is on the SQL Server Express Edition, which has a 4 GB of data limitation. If the vCenter database is full, every new transaction is written to cache. As a result when the cache is full, the database service will just stop. The cache gets cleared after a reboot of the vCenter Server, so vCenter starts to work until the cache is exhausted again.
To solve, migrate the database to the full version of the SQL Server.
If this is not an option, you may want to clean up the database using the following VMware Knowledge Base article:
Please be aware that we’re making changes which will restrict access to product updates for users without an active contract.