Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

SMBJ chunked write async #732

Open
rrj12 opened this issue Sep 2, 2022 · 4 comments
Open

SMBJ chunked write async #732

rrj12 opened this issue Sep 2, 2022 · 4 comments

Comments

@rrj12
Copy link

rrj12 commented Sep 2, 2022

I have been playing around with this library for a use case. In order to gain write/upload performance I divided my inputstream into chunks and tried to write to the SMBJ file with correct offset. I expected this change to provide good performance gain, but to my surprise the performance is same as compared to writing these chunks sequentially.

When I tried to measure the time taken for each chunks I found out that in sequential chunk write each chunk took around 1.3 secs , while the parallel chunk writes are taking upto 10-12 secs.

On further investigation I saw that the Connection.java#send is sequentialising the write requests.

     * send a packet.
     *
     * @param packet SMBPacket to send
     * @return a Future to be used to retrieve the response packet
     * @throws TransportException When a transport level error occurred
     */
    public <T extends SMB2Packet> Future<T> send(SMB2Packet packet) throws TransportException {
        lock.lock();
        Future<T> f = null;
        try {
            if (!(packet.getPacket() instanceof SMB2Cancel)) {
                int availableCredits = sequenceWindow.available();
                int grantCredits = calculateGrantedCredits(packet, availableCredits);
                if (availableCredits == 0) {
                    logger.warn(
                        "There are no credits left to send {}, will block until there are more credits available.",
                        packet.getHeader().getMessage());
                }
                long[] messageIds = sequenceWindow.get(grantCredits);
                packet.getHeader().setMessageId(messageIds[0]);
                logger.debug("Granted {} (out of {}) credits to {}", grantCredits, availableCredits, packet);
                packet.getHeader().setCreditRequest(Math
                    .max(SequenceWindow.PREFERRED_MINIMUM_CREDITS - availableCredits - grantCredits, grantCredits));

                Request request = new Request(packet.getPacket(), messageIds[0], UUID.randomUUID());
                outstandingRequests.registerOutstanding(request);
                f = request.getFuture(new CancelRequest(request, packet.getHeader().getSessionId()));
            }
            transport.write(packet);
            return f;
        } finally {
            lock.unlock();
        }
    }

Is this an expected behaviour?
Is there a way I can further improve the write performance?

File size used: 500mb
Chunk size used: 64mb
WriteApi used: File.java#writeAsync
Provider used: ArrayByteChunkProvider.java
@rrj12
Copy link
Author

rrj12 commented Sep 14, 2022

@hierynomus did you get a chance to check this? Any recommendation?

@hierynomus
Copy link
Owner

@rrj12 I've slightly changed the locks, but I don't expect that to give a humongous ROI... However have you tried configuring the async transport layer using: config.withTransportLayerFactory(new AsyncDirectTcpTransportFactory());?

@rohil-raj
Copy link

@hierynomus yes, I did configure the async transport layer in my benchmark. The performance results were inclusive of it.

@hierynomus
Copy link
Owner

Did you try with the improved locking? Did this make a difference?

hierynomus added a commit that referenced this issue Jul 3, 2023
* Ignore VSCode

* Add NtStatus.STATUS_UNSUCCESSFUL

* Add NtStatus.STATUS_INSUFF_SERVER_RESOURCES (#611)

* Add NtStatus.STATUS_IO_REPARSE_TAG_NOT_HANDLED (#514)

* Update gradle build

* Update release plugin

* Release version: 0.11.0

* Fix signing task dependency

* Release version: 0.11.1

* Use BCSecurityProvider by default for SMB3 compatibility (Fixes #638)

* Ensure DFS Path Referral times out after transactTimeout (Fixes #578)

* Only add DFSPathResolver if both client and server support DFS (#640)

* Only add DFSPathResolver if both client and server support DFS

* Fix indentation problems

* Fix incorrectly reformatted javadoc

* Format using java formatter

* One more indentation fix

* Upgrade Bouncy Castle to 1.68 to fix vulnerability report (#641)

* address issue #604 - stop closing the dfs share connection immediately. (#609)

* stop closing the dfs share connection immediately.

* Add explanatory comment

Co-authored-by: Jeroen van Erp <[email protected]>

* Add support for unregistering server from serverlist (Fixes #644) (#647)

* Add support for unregistering server from serverlist (Fixes #644)

* Fix indentation

* Reducing logging for smb3 (#650)

For each smb3 packet there's an info log message which produces a tremendous amount of output.
I would suggest to reduce log level to debug (or trace; similarly as in one of the other packet reciever classes).

* Consolidate SMBv1 error messages

* Upgrade BouncyCastle to 1.69

* Release version: 0.11.2

* Ensure artifact is signed

* Release version: 0.11.3

* Fix #665: Allow JCE KDF to work (#666)

* Fix #665: Allow JCE KDF to work

* Add header

* Add KDF unit test

* Use correct maxPayloadSize for encrypted packets (Fixes #668) (#683)

* Read fileId as long (#693)

* Read fileId of FileIdBothDirectoryInformation into a long
* Read fileId of FileIdFullDirectoryInformation into a long

* File the issue that nested folder creation throw NAME EXIST error. (#685)

* File the issue that nested folder creation throw NAME EXIST error.

* formatting

* Updated build status badges (#684)

- Added GitHub badge for Build SMBJ
- Corrected Codacy badge link
- Removed bintray badge link
- Removed Travis CI badge and configuration
- Removed Java profiler link
- Removed CircleCI configuration
- Removed unused github-ci configuration

* Use AceSize field when reading ACEs (#696)

Fixes issue seen in the wild where unnecessary padding at the end of an ACE
confused Smbj (but not Windows).

* Ensure that enough bytes are cached from InputStream to get a correct bytesLeft count for SMB2Write (fixes #669)

* GzipOutputStream integration test

* Ensure that enough bytes are cached from InputStream to get a correct bytesLeft count for SMB2Write

Co-authored-by: Stanislav Kardashov <[email protected]>
Co-authored-by: Jeroen van Erp <[email protected]>

* Add GH workflow for publishing

* Update dependencies and build file

* Rename test class to *Spec

* Release version: 0.11.5

* Use the hostname part of the TargetHint for DFS step 9 (fixes \#419) (#722)

* Slightly reduce the locking in Connection.send and DirectTcpTransport (fixes \#732)

* Fixed indentation

* Converting bytes written to long (Fixes #740)

Signed-off-by: Jeroen van Erp <[email protected]>

* Upgrading gradle to 8.0.2

Signed-off-by: Jeroen van Erp <[email protected]>

* Add Implementation manifest attributes (Fixes #743)

* Revert accidental comment of integration docker tasks

* Do not send SMB2EncryptionCapabilities NegotiationContext is !isEncry… (#752)

* Do not send SMB2EncryptionCapabilities NegotiationContext is !isEncryptionSupported (Fixes #747)

* Add test for SMB2EncryptionCapabilities

* Add preliminary changelog for new release

* Ensure we call flip() on Buffer to avoid Java8 problems (Fixes #705)

Signed-off-by: Jeroen van Erp <[email protected]>

* Ensure path is set for rmdir to prevent accidents (Fixes #756)

Signed-off-by: Jeroen van Erp <[email protected]>

* Add support for reading / writing NIO ByteBuffers (#759)

* Add support for reading / writing NIO ByteBuffers

Currently one can transfer data using streams or array, but it would be
great to have the opportunity to use NIO buffers.

This adds two new method to the File class that accept a NIO ByteBuffer.

* Implemented ByteBuffer write using ByteChunkProvider

Signed-off-by: Jeroen van Erp <[email protected]>

---------

Signed-off-by: Jeroen van Erp <[email protected]>
Co-authored-by: Christoph Läubrich <[email protected]>
Co-authored-by: Jeroen van Erp <[email protected]>

* Fix some sonatype warnings

* resolve conflict with master

* Ignore non-semver tags for release workflow

* Small warning cleanup

Signed-off-by: Jeroen van Erp <[email protected]>

* Setup ConnectionContext and AuthenticationContext for NTLM improvements

Signed-off-by: Jeroen van Erp <[email protected]>

* Refactor TargetInfo/AvPairs

* Added null check and rename field

* Refactor NtlmFunctions

* Change hierarchy of Ntlm messages

Signed-off-by: Jeroen van Erp <[email protected]>

* Next step of NTLM refactor

* NtlmNegotiate sends Domain/Workstation/Version fields

* Filter negotiateflags and use clientTargetInfo

* Rework keys in NtlmAuthenticator

Signed-off-by: Jeroen van Erp <[email protected]>

* Change to structure of NtlmAuthenticate

Signed-off-by: Jeroen van Erp <[email protected]>

* Added last changes

Required to put withIntegrity = false still, due to missing mechListMIC

Signed-off-by: Jeroen van Erp <[email protected]>

---------

Signed-off-by: Jeroen van Erp <[email protected]>
Co-authored-by: Nicholas DiPiazza <[email protected]>
Co-authored-by: ndimitry <[email protected]>
Co-authored-by: Patrick Boyd <[email protected]>
Co-authored-by: Hannes <[email protected]>
Co-authored-by: pyzhou <[email protected]>
Co-authored-by: exceptionfactory <[email protected]>
Co-authored-by: Chris Pacejo <[email protected]>
Co-authored-by: Stanislav Kardashov <[email protected]>
Co-authored-by: Stanislav Kardashov <[email protected]>
Co-authored-by: Christoph Läubrich <[email protected]>
Co-authored-by: Christoph Läubrich <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants