We really can't define them at compile time anymore as we build multiple
exes that depend on libspark in the same dir (e.g., spark api server,
pubsub, ds-publisher, webapp, etc.). That means one libspark.so (or dll)
must serve all exes. We should compile a consistent libspark to mitigate
the risk of running an exe with the wrong version (of libspark). For
example, running pubsub with a libspark configured for spark api server,
leftover from an earlier build.
To deploy to aws, connect to our aws vpn then run:
tools/deploy-docker-compose.sh
The deploy script does the following:
* builds sparkd
* builds datastream apps (pubsub, md-publisher, and webapp)
* uploads the build to aws
* bounces all services (sparkd, pubsub, md-publisher, and webapp)
Note: the name 'md-publisher' refers to DataStreamPublisher w/ MarketData
configuration
* Get rid of JsonPropertyName attributes in PubSubConfig
- The Dependency Injection framework (based around dotnet ApplicationBuilder)
does not support this attribute, and maps property names verbatim to json fields.
* Change PubSubServer json fields to match cs class property names verbatim
As far as I know, it only needs spark/protobuf config to initialize the
c++ logging system. Eventually, we should get rid of that dependency and
use builtin Microsoft Logging facilities for c# instead. In the
meantime, PubSubServer will load config from local files by default
(i.e., spark.conf).