Batched Invocation Interceptors

2 min read

2 min read

2 min read

2 min read

A batch invocation interceptor lets you flush batched requests with your own policy, and handle a failed flush.

You install an interceptor by setting the batchRequestInterceptor field of the InitializationDataInitializationDataInitializationDataInitializationDataInitializationDataInitializationDataInitializationDataInitializationDataInitializationData object you use to create your communicator. The communicator invokes the interceptor for each batch request, passing the following arguments:

  • req: an object representing the batch request being queued
  • count: the number of requests already in the queue
  • size: the number of bytes of the queued batch message, including its header

The request represented by req is not included in count and size, and is queued only when the interceptor calls enqueue on it.

The following interceptor flushes the queued requests when adding the new request would make the batch message larger than 64 KiB:

C++
constexpr int maxBatchSize = 64 * 1024; // in bytes
initData.batchRequestInterceptor =
[](const Ice::BatchRequest& req, int count, int size)
{
if (size + req.getSize() > maxBatchSize)
{
req.getProxy().ice_flushBatchRequestsAsync(
[](std::exception_ptr ex)
{
// Log the failure.
});
}
req.enqueue();
};
C#
const int maxBatchSize = 64 * 1024; // in bytes
initData.batchRequestInterceptor = (req, count, size) =>
{
if (size + req.getSize() > maxBatchSize)
{
_ = FlushAsync(req.getProxy());
}
req.enqueue();
};
static async Task FlushAsync(Ice.ObjectPrx proxy)
{
try
{
await proxy.ice_flushBatchRequestsAsync();
}
catch (Ice.LocalException exception)
{
Console.Error.WriteLine($"Failed to flush batch requests: {exception}");
}
}
Java
final int maxBatchSize = 64 * 1024; // in bytes
initData.batchRequestInterceptor = (req, count, size) -> {
if (size + req.getSize() > maxBatchSize) {
req.getProxy().ice_flushBatchRequestsAsync().exceptionally(ex -> {
System.err.println("Failed to flush batch requests: " + ex);
return null;
});
}
req.enqueue();
};
Python
maxBatchSize = 64 * 1024 # in bytes
async def logFlushFailure(flush: Awaitable[None]) -> None:
try:
await flush
except Ice.LocalException as exception:
print(f"Failed to flush batch requests: {exception}", file=sys.stderr)
def interceptor(request: Ice.BatchRequest, count: int, size: int) -> None:
if size + request.getSize() > maxBatchSize:
asyncio.create_task(logFlushFailure(request.getProxy().ice_flushBatchRequestsAsync()))
request.enqueue()
initData.batchRequestInterceptor = interceptor

Installing an interceptor does not turn off automatic flushing.