Ze Mao

how to use db session in router level unit test

4

Hi, I have added fastapi-sqlalchemy to my app and it runs well if I start from main.py, where ``` app.add_middleware(DBSessionMiddleware, db_url=settings.DATABASE_RUL) ``` is added. Now I am trying to build...

ByteBuf leak in LocalNetworkClient

1

It looks like `LocalNetworkClient` has some memory leak. @lightningrob @justinlin-linkedin It can be constantly reproduced by `testStitchGetUpdateDelete` test. ``` SLF4J: Class path contains multiple SLF4J bindings. SLF4J: Found binding in...

2020/03/31 05:25:11.737 ERROR [RequestHandlerPool] [Thread-1] [ambry-server] [] error when shutting down request handler pool {} java.lang.NullPointerException at java.io.DataInputStream.readFully(DataInputStream.java:195) at java.io.DataInputStream.readLong(DataInputStream.java:416) at com.github.ambry.network.NettyServerRequestResponseChannel.sendRequest(NettyServerRequestResponseChannel.java:41) at com.github.ambry.protocol.RequestHandler.shutdown(RequestHandler.java:64) at com.github.ambry.protocol.RequestHandlerPool.shutdown(RequestHandlerPool.java:65) at com.github.ambry.server.AmbryServer.shutdown(AmbryServer.java:305) at com.linkedin.ambry.server.AmbryLIServer.notifyShutdown(AmbryLIServer.java:66)...

Add TTL option for NettyPerfClient

3

NettyPerfClient is a Netty based tool to generate PUT/GET traffic to frontend. CLI to run this tool: ``` java -Dlog4j.configuration=file:../config/log4j.properties -cp ambry.jar com.github.ambry.tools.perf.rest.NettyPerfClient --hosts localhost --port 1174 --requestType POST --concurrency...

There is case causing NPE in GetBlobOperation. An instance of stack is as follow: ``` 2019/01/17 15:42:37.472 ERROR [NettyResponseChannel] [RequestResponseHandlerThread-0] [ambry-frontend] [] Internal error handling request /xxxxxxxx.pdf/BlobInfo with method GET...

Optimize POST's transferFrom. (transferFrom is not doing zero copy and IO size is 8K)

1

An experiment was did by sending one blob to ambry. `strace` shows that `POST` is not doing zero copy and two system call was used ('read' and 'pwrite') multiple times...

The failure: ``` com.github.ambry.router.GetBlobOperationTest > testDataChunkFailure[1] FAILED java.util.concurrent.ExecutionException: com.github.ambry.router.RouterException: Prepare for sending failed Error: UnexpectedInternalError at com.github.ambry.router.FutureResult.resultOrThrow(FutureResult.java:113) at com.github.ambry.router.FutureResult.get(FutureResult.java:99) at com.github.ambry.router.GetBlobOperationTest.doPut(GetBlobOperationTest.java:222) at com.github.ambry.router.GetBlobOperationTest.testDataChunkError(GetBlobOperationTest.java:862) at com.github.ambry.router.GetBlobOperationTest.testDataChunkFailure(GetBlobOperationTest.java:557) Caused by: com.github.ambry.router.RouterException: Prepare for...

Now PUT requests pick up partition randomly. Sometimes we want PUT request to land on a particular partition. The proposal is to add a header named `x-ambry-partition-id` and the value...

It would be convenient to have the feature that ServerAdminTool looks for the right host contains the blob. Currently, we need to do //Replicas to find out the host.

Make its default value false. Some v1 blob id are experiencing issues.