fix: nicer error message for rpc errors

[Jackmin801]

Goals:

• resolves GRPC keep alive need a nice error message #5136

[Jackmin801]

Main issue is that all RpcErrors have .details() except for https://github.com/grpc/grpc/blob/master/src/python/grpcio/grpc/_channel.py#L361-L453

I have no idea when that one is triggered though. We can catch it in the line above the RpcError catch to explicitly deal with it but I wouldn't know what information would be useful in the error message.

[codecov]

Codecov Report

Merging #5325 (4f0b7aa) into master (0d02163) will increase coverage by 1.41%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #5325      +/-   ##
==========================================
+ Coverage   85.24%   86.65%   +1.41%     
==========================================
  Files          99       99              
  Lines        6451     6454       +3     
==========================================
+ Hits         5499     5593      +94     
+ Misses        952      861      -91     

Flag	Coverage Δ
jina	86.65% <100.00%> (+1.41%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
jina/clients/base/grpc.py	85.91% <100.00%> (-2.32%)	⬇️
jina/orchestrate/deployments/config/k8s.py	100.00% <0.00%> (+0.61%)	⬆️
jina/enums.py	86.31% <0.00%> (+1.05%)	⬆️
...untimes/request_handlers/worker_request_handler.py	97.35% <0.00%> (+1.32%)	⬆️
jina/serve/runtimes/gateway/http/gateway.py	100.00% <0.00%> (+1.58%)	⬆️
jina/jaml/helper.py	83.94% <0.00%> (+2.18%)	⬆️
jina/clients/helper.py	100.00% <0.00%> (+2.38%)	⬆️
jina/orchestrate/deployments/config/helper.py	94.73% <0.00%> (+3.50%)	⬆️
...a/orchestrate/deployments/config/docker_compose.py	99.01% <0.00%> (+4.41%)	⬆️
... and 3 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[Jackmin801]

Oh and the way I reproduce the error is to run a gateway on a different computer in my network then try to ping it with the CLI:

jina ping flow grpc://192.168.100.56:12345 --attempts 10

Local pings to local gateways aren't able to produce the "keepalive watchdog timeout" error because the error is thrown when a request is mid-flight then suddenly stops replying without closing the connection. If the pinger and gateway live in the same network namespace, this is not possible as the connection status will always be known by the pinger. And thus it can never timeout.

[samsja]

lgtm, sometime the best solution only requires a couple of line of code !