Hi I observed several connection errors to my hosted slash.dgraph.io instance over the weekend:
I am using the dgraph-js lib from a node application.
Dec 12 04:08:27 PM 12-13-2020 00:08:27.360 [ERROR] 14 UNAVAILABLE: last connection error: connection error: desc = “transport: Error while dialing dial tcp 10.60.1.135:9080: i/o timeout”; last resolver error: produced zero addresses
Dec 13 08:13:04 PM 12-14-2020 04:13:04.039 [ERROR] 14 UNAVAILABLE: last connection error: connection error: desc = “transport: Error while dialing dial tcp 10.60.7.180:9080: connect: no route to host”; last resolver error: produced zero addresses
These appear to be server-generated grpc errors and I am wondering how to avoid these problems. In general, grpc unavailable errors may be handled with a retry, if applicable. I will go ahead and try to improve the resiliency of my app.
However, these do appear to be some kind of internal dns error. I am not familiar with how my instance is deployed, but I wonder if go - Handling "resolver returned no addresses" grpc error - Stack Overflow is related. Potential bug notwithstanding, is it expected that I should build in retry logic into my client-side code to handle this kind of error? Thanks for your time.