GitLab

keep dependencies up-to-date.

full changelog: https://github.com/eclipse-ee4j/grizzly/releases/tag/4.0.0-RELEASE

Acked-by: Lea Morschel
Target: master

keep external dependencies up to date

Acked-by: Albert Rossi
Acked-by: Lea Morschel
Target: master

Result:
more test coverage

Acked-by: Paul Millar
Target: master

Motivation:
Sometimes code need to perform an action before RPC request is executed,
for example:

- logging
- validation
- DoS protection / rate limitation

```
  limiter = RateLimiter.create(500);

  svc = new OncRpcSvcBuilder()
        .withServiceName("svc")
        .withCallInterceptor(c -> limiter.acquire())
        .build();
  svc.start();
```

Modification:
introduce OncRpcSvcBuilder#withCallInterceptor that injects a RPC call
consumer that is called before real work is done.

Result:
additional actions can be performed before rpc call execution.

Acked-by: Lea Morschel
Acked-by: Paul Millar
Target: master

Motivation:
when a Xdr stream is created for bigger requests the RPC layer might
require to grow the under laying grizzly buffer. As grizzly memory
manager can reallocate only buffers that ware allocated by the same
memory manager, the Xdr should know which memory manager is used by RPC
layer.

Modification:
Update Xdr to accept the corresponding memory manager. Update TPC an UDP
filters to pass memory manager used by under laying connection. Added
test to demonstrate the behavior.

Result:
Correct behavior when direct byte buffers are used.

Acked-by: Paul Millar
Target: master

Motivation:
to reduce dependency on grizzly don't use grizzly specific Buffer
without need.

Modification:
Update RpcAuthGss to expose packet header as java.nio.ByteBuffer

Result:
less dependency on grizzly

Acked-by: Paul Millar
Acked-by: Lea Morschel
Target: master

Motivation:
In some situations client or server wants to now if TLS in applied on
the connection, for example before sending the data over the wire or
for data protection validation.

Modification:
introduced  RpcTransport#isTLS that checks existence if SSLFilter in
the grizzly pipeline.

Result:
simple way to check TLS status of a connection.

target: master
Acked-by: Lea Morschel

Motivation:
according to the current draft on startTLS client should send an empty
verifier, while server must reply with "STARTTLS" string.

See: https://www.ietf.org/archive/id/draft-ietf-nfsv4-rpc-tls-11.txt

Modification:
Update RpcAuthTypeTls to have a different behavior in the client and
server modes.

Result:
Interoperability with Solaris server.

Acked-by: Lea Morschel
Target: master

time to mover forward

Acked-by: Lea Morschel
Target: master

Motivation:
The StartTlsFilter holds a state to indicates that TLS handshake should
be started. This state currently shared by all connections forcing all
following connections to start TLS handshake, even if not requested.

Modification:
Use Connection object's attribute store to hold connection specific
flag. As we only interested in a on/off flag a dummy object is used as a
value.

Result:
An earlier tls session has no influence on a following one.

Fixes: #95
Acked-by: Paul Millar
Target: master

Motivation:
use up-to-date release with bugfixes and enhancements.

Modification:
use grizzly-3.0.0

Result:
up-to-date dependency. no visible changes

Acked-by: Lea Morschel
Target: master

build(deps-dev): bump bcprov-ext-jdk15on from 1.60 to 1.67

Bumps [bcprov-ext-jdk15on](https://github.com/bcgit/bc-java) from 1.60 to 1.67.
- [Release notes](https://github.com/bcgit/bc-java/releases)
- [Changelog](https://github.com/bcgit/bc-java/blob/master/docs/releasenotes.html)
- [Commits](https://github.com/bcgit/bc-java/commits

)

---
updated-dependencies:
- dependency-name: org.bouncycastle:bcprov-ext-jdk15on
  dependency-type: direct:development
...
Signed-off-by: dependabot[bot] <support@github.com>

Motivation:
The API to start TLS handshake is not handy as requires an extra RPC
NULL call.

Modification:
introduce RpcCall#startTLS to encapsulate the RPC NULL call.

Result:
a slightly better API

Acked-by: Lea Morschel
Target: master

assertFalse(!value) => assertTrue(value)
assertTrue(a.equals(b)) -> assertEquals(a, b)

Acked-by: Marina Sahakyan
Acked-by: Lea Morschel
Target: master

Motivation:
Test code doesn't belong to the main library.

Modification:
remove SimpleRpcClient and SimpleRpcServer. Add corresponding benchmark
that provides the same functionality.

Result:
no test code in the core library.

Acked-by: Lea Morschel
Acked-by: Paul Millar
Target: master

fixes falsely places package

Acked-by: Lea Morschel
Target: master

Acked-by:
Target: master

Motivation:
As OncRpcClient implements AutoCloseable interface, using in
try-with-resource block is more natural than try-finally

Modification:
update OncRpcSvc#publishToPortmap and OncRpcSvc#clearPortmap to make use
of try-with-resource construction.

Result:
more simple and up-to-date logic

Acked-by: Lea Morschel
Acked-by: Albert Rossi
Acked-by: Paul Millar
Target: master

no logic changes

Target: master

Motivation:
The RPC messages might have different sizes and purposes. Some of
them and small, others are big, for example nfs GETATTR call/reply
are ~128 byes, but IO requests like READ and WRITE might be 1MB big.
Moreover, the data used by IO requests are often read from or written
to disk through java NIO API. Thus direct byte buffers are preferred.

Modification:
Add a possibility to control the memory allocation policy. Update
OncRpcSvcBuilder to expose new functionality, by preserving default
behavior.

Result:
More flexible memory management with rule of thumb:

For small requests use HEAP allocator. In other cases POOLED
has a better performance.

AllocatorBenchmark.allocate           heap        128  thrpt   20  49791239.209 ± 122939.377  ops/s
AllocatorBenchmark.allocate             bb        128  thrpt   20  31980902.534 ± 130232.954  ops/s
AllocatorBenchmark.allocate  pooled-direct      16384  thrpt   20  18150104.447 ±  94087.140  ops/s
AllocatorBenchmark.allocate  pooled-direct        128  thrpt   20  18083772.708 ± 231314.463  ops/s
AllocatorBenchmark.allocate  pooled-direct       1024  thrpt   20  18051147.691 ±  75325.348  ops/s
AllocatorBenchmark.allocate         pooled      16384  thrpt   20  12510167.658 ± 189768.727  ops/s
AllocatorBenchmark.allocate         pooled        128  thrpt   20  11313056.455 ± 101182.275  ops/s
AllocatorBenchmark.allocate         pooled       1024  thrpt   20  10704443.770 ± 417986.884  ops/s
AllocatorBenchmark.allocate           heap       1024  thrpt   20  10039347.620 ±  21660.670  ops/s
AllocatorBenchmark.allocate             bb       1024  thrpt   20   9731756.363 ± 111935.133  ops/s
AllocatorBenchmark.allocate      bb-direct        128  thrpt   20   1167915.132 ± 303779.958  ops/s
AllocatorBenchmark.allocate             bb      16384  thrpt   20   1103752.015 ±   4981.959  ops/s
AllocatorBenchmark.allocate           heap      16384  thrpt   20   1096764.395 ±   3383.219  ops/s
AllocatorBenchmark.allocate      bb-direct       1024  thrpt   20    942751.899 ±  63066.509  ops/s
AllocatorBenchmark.allocate  pooled-direct    1048576  thrpt   20    806817.161 ±  15408.086  ops/s
AllocatorBenchmark.allocate         pooled    1048576  thrpt   20    690660.597 ±   6888.734  ops/s
AllocatorBenchmark.allocate      bb-direct      16384  thrpt   20    488973.714 ±   6925.715  ops/s
AllocatorBenchmark.allocate      bb-direct    1048576  thrpt   20     15487.488 ±     70.131  ops/s
AllocatorBenchmark.allocate           heap    1048576  thrpt   20     14889.964 ±    113.748  ops/s
AllocatorBenchmark.allocate             bb    1048576  thrpt   20     14884.161 ±    142.970  ops/s

Target: master
Acked-by: Albert Rossi
Acked-by: Paul MiLlar

Acked-by: Lea Morschel
Target: master

Log the address we disconnected from in the exception

Acked-by: Paul Millar
Acked-by: Lea Morschel
Target: master

time to move forward...

Acked-by: Lea Morschel
Acked-by: Albert Rossi
Target: master

Acked-by: Lea Morschel
Target: master

build(deps): bump junit from 4.12 to 4.13.1

Bumps [junit](https://github.com/junit-team/junit4) from 4.12 to 4.13.1.
- [Release notes](https://github.com/junit-team/junit4/releases)
- [Changelog](https://github.com/junit-team/junit4/blob/main/doc/ReleaseNotes4.12.md)
- [Commits](https://github.com/junit-team/junit4/compare/r4.12...r4.13.1

)
Signed-off-by: dependabot[bot] <support@github.com>

build(deps): bump guava from 24.1-jre to 24.1.1-jre

Bumps [guava](https://github.com/google/guava) from 24.1-jre to 24.1.1-jre.
- [Release notes](https://github.com/google/guava/releases)
- [Commits](https://github.com/google/guava/commits

)
Signed-off-by: dependabot[bot] <support@github.com>

Motivation:
improve test coverage

Acked-by: Albert Rossi
Target: master

Target: master

[SECURITY] Use HTTPS to resolve dependencies in Maven Build

This fixes a security vulnerability in this project where the `pom.xml`
files were configuring Maven to resolve dependencies over HTTP instead of
HTTPS.
Signed-off-by: Jonathan Leitschuh <Jonathan.Leitschuh@gmail.com>

Acked-by: Paul Millar
Acked-by: Svenja Meyer
Target: master

Acked-by: Paul Millar
Target: master

Motivation:

RpcCall#call(proc, args, callback, auth) method throws EOFException
when no callback is provided and client is disconnected. However,
there is no guarantee, that after call is complete client is still
able to send the request. IOW, it client code depends on message
delivery, then callback have to be provided.

Modification:
update RpcCall#call not to check connection status.

Result:
more predictable client behaviour. NOTICE, that current code, which
expect such functionality is now broken, but it broken by design anyway.

Acked-by: Paul Millar
Target: master