I have following (simplified) steps in pipeline:
- Read agent IP;
- Add agent ip to azure cosmosdb;
- Run tests which read data from cosmosdb.
Today, for first time for like 1 year + I received error that in step 3 IP from which request to azure comosdb was orignated is different than ip read in step 1. All above steps are under same job (so same agent is used). How it is possible that agent changed it ip during single agent runtime? I think that should be reported to MS but first wanted to know if anyone else thinks/knows otherwise.