We will be releasing a software update next Thursday (8/6/2020) to the network per the release schedule found in the handbook. This update will include a significant improvement in how log uploading is achieved using CloudWatch, as well as a series of stability improvements. It also has a restructuring of the comms code to allow it to be better shared between Elixxir and Praxxis.
This release also includes an updated node handbook, which can be found here.
This update will go out over Auto Update on 8/6/2020, you only need to update manually if that is disabled. (Auto Update is enabled by default)
Implemented Gateway IP configuration for overriding gateway IP and special support for gateway on the same machine as the server
Support for comms split
Updated internal dependencies
Improved testing
Hacky fix for memory leak when gpu is enabled. Underlying cause has not been determined, but threads which were not closing now have timeouts
Increased size of outbound databuffer from computation to ensure it is not blocked on transmission
Future Work
We have been working hard on the next steps for the network, which are the re-release of the xx messenger and the launch of xx consensus.
The Elixxir and Praxxis teams worked together on the comms split and have continued to work together to implement a genetic implementation of the Gossip protocol within the shared comms repo. This will be used by Elixxir to build support for the xx messenger by having gateways gossip various information. We will release a blog post with more details about the gossip protocol next week.
We have also been rewriting the cryptographic storage for ArrowSDK in preparation for getting it ready for xx messenger running on the BetaNet.
Over the next few weeks we hope to turn a more detailed focus on the network and improving its stability and performance.
On 06-Aug-20 18:18:17 there was a gateway wrapper log entry: An error occurred (413) when calling the PutLogEvents operation:
Since then, the log gets a new entry every 1-30 seconds: [ERROR] 10-Aug-20 18:38:35: Parameter validation failed: Invalid type for parameter sequenceToken, value: None, type: <class 'NoneType'>, valid types: <class 'str'>
Is there something to manually update on my side, or OK to ignore this for now?
I was able to reproduce the error An error occurred (InvalidParameterException) when calling the PutLogEvents operation: Log event too large: 1395334 bytes exceeds limit of 262144
You can reduce the logLevel in the yaml files. But if you insist on logLevel higher than INFO then you can disable uploading logs with --disable-cloudwatch in the node and gateway service files. DEBUG and TRACE data does not need to be uploaded to AWS.