Release notes

Release notes

Couchbase Server 3.0 is a major release that extends Couchbase lead as the most performant and scalable NoSQL database for mission critical enterprise applications. In addition to adding great new functionality for the enterprise, 3.0 delivers many new features that make it easier for developers and administrators to work with Couchbase Server, making it the best choice for your NoSQL projects.

In addition to adding great new functionality for the enterprise, 3.0 delivers many new features that make it easier for developers and administrators to work with Couchbase Server, making it the best choice for your NoSQL projects.

Release notes for 3.1.6

Released in July 2016, Couchbase Server 3.1.6 is a maintenance release in the 3.x series for Couchbase Server. This maintenance release has several fixes related to DCP, view engine, and XDCR, as well as improved mutex handling to correct race issues.

Fixed issues in 3.1.6

Here are the notable fixes in the 3.1.6 release:

Issue Description
MB-20105 Purge Seqno would be reset in case there was a compaction call that was not purging any items, causing tombstone items to not be purged.
MB-20054 Memcached aborted while buckets were deleted.
MB-19982 Memcached was timing out due to a lock inversion issue between DCP and the connection map.
MB-19843 View engine would fail with DCP start sequence number greater than end-sequence number error, and then fail to roll back.
MB-19832 XDCR would start with an infinite loop when a node that was previously removed from a cluster joined the cluster back.
MB-19752 Using the command cbreset_password to reset the admin password would remove the read-only user’s account.
MB-19732

MB-19255

On Windows, rebalance would exit because of bad replicators.
MB-19706 An online upgrade from Couchbase Server version 2.5 to the version 4.5 failed.
MB-19673 The potentially misleading DCP producer end-stream log message was improved.
MB-19636 An offline upgrade from Couchbase Server version 2.5 to any version of a Couchbase Server release 3.x could result in a potential data loss.
MB-19503 Under low mutation rate, long pauses could occur between creating and closing a DCP stream.
MB-19319 Executing log collection from the UI through cbcollect_info would hang on SUSE11.
MB-19278 Potential deadlock can occur in DCP due to the lock-order inversion
MB-19113

MB-19380

MB-19259

Several fixes have been introduced to improve mutex handling and address potential race conditions across the data service.
MB-16366 A race between the front-end memcached thread and DCP snapshotting would trigger an assert.

For a full list of issues fixed in 3.1.6, see the following JIRA item .

Release notes for 3.1.5

Couchbase Server 3.1.5, which is released in April 2016, is a maintenance release in the 3.x series for Couchbase Server. This release supersedes Couchbase Server version 3.1.4.

Fixed issues in 3.1.5

The following issues are now fixed:

Issue Description
MB-19093 Data service (memcached) can lock up when replica indexes for Views are enabled.
MB-17506 Improvement: New private setting that allows disabling the cluster map being included in a Not My VBucket response from the server to the client. Which can in some cases reduce network bandwidth.
MB-17279 If DCP connection is closed without waiting for each replication stream to be closed, a race condition can cause KV engine to pick up the connection after some time, and cause memcached to crash.
MB-17241 Stale DCP stream are not picked up during a DCP takeover operations, causing rebalance to become stuck.
MB-17132 When a node is added to a cluster, a new UUID is generated for the node.
MB-17026 If memcached crashes, the view-engine continuously retries to get the current high sequence number. The following log message can be seen in the view-engine log file “dcp all seqs timed out after 2.0 seconds. Waiting...”.
MB-17006 During a backfill operation, if a front-end mutation is on-going, a race condition in DCP can cause a mutation item to remain stuck in the DCP checkpoint queue without getting transmitted.
MB-16836 Cbstats reset command does not reset ep-engine background fetch counter ( ep_bg_fetched ).
MB-16752 Memcached process can crash if it runs out of file descriptors during log rotation.
MB-16656 Incorrect ep-engine sequence number can cause view-engine operations to timeout.
MB-16632 Minimized contention between store operations and DCP connection notifications can impact operational latency.
MB-16357 If a vBucket state changes from active to replica while performing compaction, a race condition between the compaction thread and memcached thread can cause an assertion and trigger a crash.

Release notes for 3.1.4

Due to a bug affecting system stability when using replica view indexes ( MB-19093 ), the 3.1.4 release is superseded by 3.1.5.

Fixed issues in 3.1.4

The following issues have been fixed:

Issue Description
MB-17279 If DCP connection is closed without waiting for each replication stream to be closed, a race condition can cause KV engine to pick up the connection after some time, and cause memcached to crash.
MB-17241 Stale DCP stream are not picked up during a DCP takeover operations, causing rebalance to become stuck.
MB-17132 When a node is added to a cluster, a new UUID is generated for the node.
MB-17026 If memcached crashes, the view-engine continuously retries to get the current high sequence number. The following log message can be seen in the view-engine log file “dcp all seqs timed out after 2.0 seconds. Waiting...”.
MB-17006 During a backfill operation, if a front-end mutation is on-going, a race condition in DCP can cause a mutation item to remain stuck in the DCP checkpoint queue without getting transmitted.
MB-16836 Cbstats reset command does not reset ep-engine background fetch counter ( ep_bg_fetched ).
MB-16752 Memcached process can crash if it runs out of file descriptors during log rotation.
MB-16632 Minimized contention between store operations and DCP connection notifications can impact operational latency.
MB-16357 If a vBucket state changes from active to replica while performing compaction, a race condition between the compaction thread and memcached thread can cause an assertion and trigger a crash.

Release notes for 3.1.3

Couchbase Server 3.1.3, which is released in December 2015, is the third maintenance release in the 3.x series for Couchbase Server.

Fixed issues in 3.1.3

The following issue has been fixed:

Issue Description
MB-16915 Memcached crash caused by a race condition when a DCP producer is closed while a backfill task is running. This can happen when rebalance is terminated or when XDCR connections are terminated.

Release notes for 3.1.2

Couchbase Server 3.1.2, which is released in November 2015, is the second maintenance release of Couchbase Server 3.1. This release fortifies Couchbase Server and includes several bug fixes that improve the stability, and security of the product for our customers.

Known issues in 3.1.2

The following issues are targeted to be fixed in a future release:
Issue Description
MB-16689 Memcached process can crash if it runs out of file descriptors during log rotation.

Workaround: Ensure that memcached has sufficient file descriptors. For a normally packaged installation, the maximum number of file descriptors is automatically configured. If performing a non-root installation, ensure to set the maximum number of file descriptors correctly.

Fixed issues in 3.1.2

The following issues have been fixed:

Issue Description
MB-16385 Querying a view based on a subset of partitions can result in a massive memory usage.
MB-16357 If a vBucket state changes from active to replica while performing compaction, the race condition between the compaction thread and memcached thread can cause an assertion and trigger a crash.
MB-16421 If a getMeta is issued at the destination cluster during XDCR followed by a GET request by the client, the background fetch operation for the item does not complete and causes a large number of disk reads and client side timeouts.
MB-16528 If a delta-node recovery is started after the bucket configuration is updated but before the bucket is loaded into memcached, a rebalance operation might eject the node from the cluster, and the cluster vBucket map might still contain the node.
MB-13948 The mapping phase of the view MapReduce operation takes a lot of memory if there are lots of key-value pairs emitted per document.
MB-16055 Changing a cluster RAM quota using CLI removes the read-only user.
For a full list of issues fixed in 3.1.2, see the following JIRA item.

Release notes for 3.1.1

Couchbase Server 3.1.1, released in October 2015, is a first maintenance release for Couchbase Server 3.1. It further fortifies Couchbase Server 3.1.0 and includes some critical bug fixes.

Fixed issues in 3.1.1

The following issues have been fixed:

Issue Description
MB-16245 [couchbase-cli] The arguments --email-port and --email-password for setting-alert did not work in Couchbase Server version 3.0.3 and 3.1.
MB-16207 Corrupted SASL packet caused memcached to crash.
MB-16178 Improve the reliability of the TAP to DCP conversion rebalance when upgrading from the version 2.5.x.
MB-16159 DCP consumer would consistently take 6 seconds to acknowledge a 20Mb mutation.
MB-16131 Improve robustness of passive DCP streams.
MB-16125 Memcached would hang during the shutdown.
MB-16064 Rebalance is failing when one node was unavailable.
MB-16045 DCP packets would trigger assertion in item.h:373 .
MB-16042 DCP packets would trigger assertion in ep-engine-vbucket.h:225 .
MB-16013 Version 2 of XDCR would consume a lot of memory with large documents.
MB-15972 Rebalance would repeatedly fail when trying to add two additional nodes to a two-node cluster while a client was making view requests.
MB-15926 Race condition in the Cluster manager would cause memcached to crash.
MB-15914 Race condition that can delete vbucket files.
MB-15837 A new vbucket UUID was not generated during TAP takeover leading to potential data loss in the XDCR replication.
MB-15826 [ cbbackup ] Incremental backup was growing from MB to many GB.
MB-14845 Support for encryption cipher suites with key-length below 128 bits is removed.

Release notes for 3.1.0

Couchbase Server 3.1.0, released in July 2015, is a minor update release for Couchbase Server 3.0. It includes some performance enhancements and important bug fixes. This release also extends Couchbase Server to support SuSE Enterprise Linux 11.

For Couchbase Mobile users, Couchbase Server 3.1.0 is production certified with Couchbase Sync Gateway 1.1. Sync Gateway users are encouraged to upgrade to 3.1 to take advantage of the many improvements since the 2.5.2 release.

Enhancements

The following enhancements have been introduced in Couchbase Server 3.1.0:

  • Support was added for SuSE Enterprise Linux 11.
  • There are major performance improvements for Map-Reduce views.
  • Support was added for the Get Sequence Numbers command.
  • Users can now specify the SSL ciphers list.

Fixed issues in 3.1.0

The following issues have been fixed:

Issue Description
DOC-653 Mac OS X automatic updates from Sparkle: If you have Couchbase Server 3.1.0 or later installed, you will have an option to automatically update the version of Sparkle. Use Check for updates to install the new Sparkle release.
MB-12319 Mac OS X 64-bit: Upgrade with Check for Updates didn't work due to the missing signature.
MB-12359 Disable SSLv2/SSLv3 for memcached ports (POODLE vulnerability).
MB-12451 A mcd leak was observed when a socket closes during read.
MB-13142 The tool cbdocloader would fail when a memcached bucket was present.
MB-13160 View emitting only one field of a document results in multiple rows with different values.
MB-13503 The ns_connection_pool may never wake up the processes waiting for a connection.
MB-14068 Partition versions contain duplicates.
MB-14288 Value size limit is not configurable.
MB-14305 The ns_server fails to detect lost replica building stream. Rebalancing was slow when adding 3 nodes to the cluster.
MB-14537 Due to race condition per active vBucket on a failed-over node, if the wrong I/O thread wins it could cause that the most recent disk flush commit gets rolled back. Therefore, some of the vBuckets can lose the items in the replication queue and also can additionally lose the most recent disk flush commits.
MB-14545 Implementation of the DCP command to get all sequence numbers.
MB-14774 Add --from-date and --to-date option to cbrestorewrapper .
MB-14833 Fixed the performance and high memory consumption issues with the cbackup tool.
MB-14844 Disabled SSLv3 processing for TLS records and, as a result, if the last byte contains a correct value the entire padding is considered acceptable.
MB-14845 Not permitting “low” encryption cipher suites; currently, Couchbase Server supports 128bit or higher encryption algorithms.
MB-14904 Performance of view responses in Sync Gateway has been improved.
MB-14924 It was possible for a user to crash the server by sending an invalid snapshot_marker packet.
MB-14976 Support was removed for SSL v3 in the memcached SSL server sockets to mitigate against the POODLE attack.
MB-15049 Tree building in views needed to be fixed with a complex reduce.
MB-15069 Fixed LOGJAM discovered in SSL/TLS, which allows an attacker to downgrade a connection to weak export-grade crypto and potentially read/modify data.
MB-15206 Large items could potentially use up a lot of memory if left in the replica's checkpoint queues.
MB-15213 When the DCP Flow Control buffer was set to zero, the Producer would enable flow control in an erroneous way.
MB-15292 Fixed the issue with CouchbaseAtomic::Load() in memcached.
MB-15321 Compaction would break if the os_mon application terminated for some reason.
MB-15353 Request to properly find the latest rollback point was resolved.
MB-15472 Spatial views don't work on a cluster.

Release notes for 3.0.3

Couchbase Server 3.0.3, released in March 2015, is the third maintenance release for Couchbase Server 3.0. This release further fortifies Couchbase Server 3.0 and includes some critical bug fixes.

Fixed issues in 3.0.3

The following issues are fixed in the Couchbase Server 3.0.3 release.

Issue Description
MB-7250 The Mac OS X version of Couchbase Server was not signed by a valid developer key causing users to use an additional step to launch the application. Now, when Couchbase Server is installed on Mac OS X, the signed developer key is available, and no additional steps are required.
MB-10370 Under the DGM scenario and a write-heavy workload, the system would reach the bucket quota quickly and did not recover.
MB-11527 Saturated disk I/O would cause compaction to fail and then crash and fill the disk. This issue was fixed by changing the compaction_write_queue_cap engine parameter at runtime.
MB-12849 The /etc/init.d/couchbase-server script to manage starting/stopping/status of Couchbase Server on CentOS was failing on CentOS 6.6.
MB-12967 The Unicode collation in Couchbase Server had an error that resulted in wrong results.
MB-13167 When a Couchbase cluster is under upgrade, the cluster is in a mixed state of TAP/DCP and, in such situation, the rebalance fails.
MB-13205 Under the DGM scenario and a write-heavy workload, the system reaches the bucket quota quickly, and the bucket reaches hard OOM with full ejection.
MB-13284 There was a possibility of a socket leak in the xmem connection pool for XDCR.
MB-13286 The rebalance would get hung when a producer did not send a no-op to the consumers.
MB-13359 DCP was getting stuck in the intracluster replication queue when the checkpoint_interval was set to 60.
MB-13386 DCP backfill task was incorrectly reading sequence numbers causing the connection to be hung.
MB-13458 While running cbbackup , a segmentation error would cause the backup to fail.
MB-13470 Installation as a non-root user was broken in Couchbase Server version 3.0.2.
MB-13479 Take purging into account when telling DCP clients to roll back.
MB-13757 [Upgrade 2.5.1->3.0.2] Rebalance was failing with an error bad replicas as replication would fail to start.
MB-14003 Upgrading from 2.5.2 to 3.0.2 MP3 in a DGM situation causes a failed assertion [bySeqno >= 0] .

Release notes for 3.0.2

Couchbase Server 3.0.2, released in December 2014, is the second maintenance release for Couchbase Server 3.0. This release further fortifies Couchbase Server 3.0 and includes some critical bug fixes in the areas of database engine, performance, security, and server operations. It also includes several bug fixes related to the Windows platform.

Fixed issues in 3.0.2

The following issues are fixed in the Couchbase Server 3.0.2 release.

Issue Description
Couchbase bucket
MB-12160 When a key is locked, setWithMeta() is able to update the key even when the set fails.
MB-12305 memcached was crashing during the online upgrade from the Couchbase Server version 2.5.0 to version 3.0.1.
MB-12647 gethrtime() implementation on Windows can cause requests to store() with CAS to be incorrectly processed.
Performance
MB-12117 Access log locks for heavy write workloads can cause performance slowdowns.
MB-12576 Massive increase in disk write queue size on both SSD and HDD was detected.
Security
MB-8872 A number of CAPI REST API endpoints is not secured.
MB-9890 XDCR crash logs can sometimes have document contents.
MB-10262 Corrupted key in a data file rolls backwards to an earlier version, or disappears without detection.
MB-12288 The XDCR-over-SSL proxy allows for a connection to any port on the destination nodes.
MB-12358 Remove support for SSL v3 in ns_server SSL server sockets to mitigate against the POODLE attack.
MB-12655 Configuration replication of certain per-node keys is broken after a node is upgraded offline from 2.x to 3.0 and then added back to formerly 2.x cluster that is now upgraded to 3.0.
MB-12695 Certain XDCR logging exposes remote cluster passwords.
Server operations
MB-11935 Failed connections on port 11213 cause Unable to listen errors and server restarts.
MB-12156 Race condition with time check when changing the data path may lead to deletion of all buckets after adding a node to a cluster.
MB-12382 Offline cluster upgrade failed when upgrading from 2.0.1-170-rel to 3.0.1-1422-rel.
MB-12706 In an XDCR environment, when add-delete-add operations are performed, the second add operation fails when performed on temporary items.
Windows
MB-11611 memcached.logs can grow large after the node failure and bucket flush.
MB-12091 Compact files might not be cleaned up after compaction.
MB-12247 View engine failed to index documents in a query after the rebalance.
MB-12371 Incremental backup must not consume additional space with no delta.
MB-12483 Online upgrade from 2.0.0 to 3.0.1 with a single bucket might fail warmup.
MB-12238 An infinite timeout on the outgoing xmem request might lead to XDCR hanging when there are network or NAT issues.

Known issues in 3.0.2

The following known issues are associated with Couchbase Server release 3.0.2.

Issue Description
MB-12359 Remove support for SSL v3 in memcached SSL server sockets to mitigate against the POODLE attack.
Note: Port scanners might detect that port 11207 negotiates SSLv3. Upgrading to the latest 2.x client SDKs blocks pre-TLS communication and reduces the risk of the POODLE bug.

Release notes for 3.0.1

Couchbase Server 3.0.1, released in October 2014, is the first maintenance release for Couchbase Server 3.0. This release focuses on getting the system stability bugs fixed for the Windows platform and providing general availability. In addition, this release includes some critical bug fixes related to meta data corruption.

Changes in functionality

Support was added for multidimensional Geospatial indexes.

Note: This feature is currently only a development preview.

Fixed issues in 3.0.1

The following issues are fixed in the Couchbase Server 3.0.1 release.

Issue Description
Couchbase bucket
MB-12197 Bucket deletion failing with error 500 and reason: unknown {"_":"Bucket deletion not yet complete, but will continue."}
MB-11955 Could not recreate the default bucket after deleting it.
MB-11934 After warmup, fewer items are shown.
MB-11804 Memcached error #132 Internal error: Internal error for vbucket... when set key to bucket.
Installation
MB-12321 New message box is thrown during the silent installation.
MB-12135 Once installation of the Couchbase Server is completed, it launches a console page that opens with a blank page.
MB-11567 Failed to load Beer Sample in the initial setup.
Rebalance
MB-12328 Item flags are mangled during rebalance.
MB-11948 Simple test broken - Rebalance exited with the following reason: unexpected_exit..{dcp_wait_for_data_move_failed,"default",254,..wrong_rebalancer_pid}}}
Upgrade
MB-12425 Bucket failed to restore after an offline upgrade from 2.0.0 to 3.0.1.
MB-12209 The offline upgrade from 2.5.x to 3.0.1 failed.
MB-11930 All data was lost after an offline upgrade from 2.0.0 to 3.0.0.
Views
MB-12138 View query fails with the error 500 and reason: reason {"error":"error","reason":"{index_builder_exit,89,<<>>}"}

Known issues in 3.0.1

The following issues are associated with Couchbase Server release 3.0.1.

Issue Description
Database operations
MB-12647 In situations with high throughput (when the client is making multiple parallel requests), or with very low network latency (such as when running on localhost ), there is a high probability that requests to store() with CAS might incorrectly be processed on the Windows Server due to the change in gethrtime() implementation. It is recommended that Windows users stay on version 2.5.x, or wait for the next maintenance release 3.0.2.
Performance
MB-12353 Regression (from 2.5.1) in rebalance out with views and query workload.
MB-12349 Regression (from 2.5.1) in rebalance after failover with views and query workload.
MB-12293 Regression in 4 to 3 node rebalance out on the Windows 2012 Server.
Upgrade
MB-12655 Replication of certain per-node keys is broken after a node is first upgraded offline from 2.x to 3.0 and then added back to the formerly 2.x cluster that is now upgraded to 3.0.

Workaround: When performing an online upgrade from Couchbase Server 2.x to 3.x, always fully delete the 2.x package (including the config files) before installing the 3.x package.

MB-12483 Data lost with online upgrade from 2.0.0 to 3.0.1 and cluster reboot.
Backup
MB-14833 Couchbase Server Community Edition 3.0/3.0.1 : Customers have reported issues with the slow runtime and high memory consumption using the cbbackup tool.

If you are trying to back up a large dataset, you might encounter performance regressions or high memory consumption with the CBBackup tool.

Workaround : To resolve this issue:
  1. Download the cbbackupwrapper patch from Amazon Simple Storage Service (Amazon S3):
    > wget https://s3.amazonaws.com/support.couchbase.com/CBBackup-Patch/orig.patch
  2. Apply the patch to the file /opt/couchbase/lib/python/cbbackupwrapper :
    >patch -p0 < orig.patch >
  3. This patch enables the --sequential flag, which controls the limits of the number of backfills during the backup. A sample command looks as follows:
    > /opt/couchbase/bin/cbbackupwrapper http://<cluster-ip>:8091 <backup_dir> 
    -u <username> -p <password> --path /opt/couchbase/bin/ 
    --sequential -n <no_vbucket_per_batch>
    Note: Pay attention to the number you are passing to the -n flag, which controls the number of vbuckets to back-fill at a time. Tune it according to the memory headroom you have on your Couchbase node. By default, the batch size is 100.
Note: Couchbase is working on known issues related to performance and upgrading on the Windows platform. When upgrading to the Couchbase Server Windows 3.0.1 release, test your applications extensively before upgrading. Couchbase also requires that you upgrade to the latest available production version before upgrading to Couchbase Server 3.0.1. For Enterprise Edition customers, that version is Couchbase Server 2.5.1. For Community Edition users, that version is Couchbase Server 2.2.0.

Release notes for 3.0.0

Couchbase Server 3.0, released in September 2014, includes many major new enhancements and features:

  • Performance and scale

    This feature allows you to run at planet scale with more data, faster I/O, and improved indexing.

    • Tunable memory lets you configure memory to optimize meta data, data caching and eviction strategies, allowing configurations ranging from full in-memory database to large disk-based datasets.
    • Shared thread pool per node delivers high I/O throughput and low latency for data reads, writes, and other operations such as database warm-up.
    • Stream-based views based on Database Change Protocol (DCP), a new streaming protocol, reduce latency for view updates, and provide faster view consistency and fresher data.

    For more information, see:

  • Ultra-high availability

    This feature provides highly available applications with faster and easier data replication across multiple data centers, and advanced disaster recovery options.

    • Stream-based XDCR (Cross Data Center Replication) lowers XDCR latency using memory-to-memory replication of data from the source cluster to the destination cluster, even before it reaches the disk.
    • Delta-node recovery permits that nodes be quickly added back after they are repaired, so they can catch up incrementally from where they left off.
    • Incremental backup and restore enables administrators to quickly back up only modified data in the database, making the backup more efficient for larger datasets.

    For more information, see:

  • Improved security

    This feature provides secure on-the-wire data access for applications and administrators.

    • Encrypted admin access enables HTTPS access for the Couchbase admin console and the REST API.
    • Encrypted data access with SSL encrypts data in-flight between client and server.

    For more information, see:

  • Simplified administration

    This feature provides easy cluster management and resource governance.

    • Pause and resume XDCR enables XDCR streams between source and destination clusters to be paused during maintenance windows and resumed later, continuing from where they left off.
    • Graceful failover enables nodes to be safely removed from the cluster after all in-flight operations are completed.
    • Bucket priority enables prioritization of buckets so that additional I/O resources can be assigned to the higher priority buckets for faster workload execution.
    • Cluster-wide diagnostics is provided for improved cluster management.

    For more information, see:

  • Developer empowerment

    This feature allows you to build richer and more powerful applications. It provides:

    • Fully integrated native JSON, so that your application can work with high level objects that can be automatically serialized to JSON through the Couchbase APIs.
    • Powerful async and reactive interfaces in Java and .NET, which enable building of richer applications with better IO performance.
    • Rich framework integrations with open source frameworks such as Spring, gEvent, twisted and Ottoman.

Couchbase .NET incompatibility

An incompatibility between Microsoft Framework 4.0 or earlier, the Couchbase .NET SDK 1.3.7 or earlier, and Couchbase Server 3.0 has been identified where the URL generated for a View operation on the Couchbase bucket is improperly encoded. This causes view requests to fail with the following message in the body:

"Design document not found, body: {"error":"not_found","reason":"missing"}"
Impact
The affected Microsoft .NET versions are 4.0 or earlier, Couchbase .NET SDK 1.3.7 or earlier, and Couchbase Server 3.0 (including Beta) is >= 3.0.
Microsoft .NET version 4.5 changes the way Uri.ToString handles its Unicode encoding, which resolves the issue. In this case, clients running on this version or higher of the CLR are not impacted. Additionally, Couchbase Server 2.5 and lower are not impacted.
Note: Couchbase .NET 1.3.8 has a patch that resolves the issue for all versions of Couchbase Server and Microsoft .NET Frameworks.
Workaround
In addition to the patch released with Couchbase .NET SDK version 1.3.8, add the following element to your App.Config or Web.Config to enable support for all SDK and Server versions:
<uri>
    <iriParsing enabled="true"/>
<uri>
            

Customers who are not ready or cannot upgrade their Couchbase .NET SDKs to 1.3.8 or greater are strongly advised to add this element to their deployments if they intend to use the Couchbase .NET SDK with Couchbase Server 3.0 or greater.

References
https://connect.microsoft.com/VisualStudio/feedback/details/758479/system-uri-tostring-behaviour-change

Views behavior change

The stale=false view query argument has been enhanced. It considers all document changes that have been received at the time the query was received. This means it is no longer needed to use the durability requirements or the Observe feature to block for persistence in application code before issuing the stale=false stale query. It is recommended that you remove all such application level checks after completing the upgrade to the 3.0 release.

Fixed issues in 3.0.0

There are many bugs fixed in the 3.0.0 release. Some of the notable fixes include the following:

Issue Description
Couchbase bucket
MB-10059 Replica vbucket simply ignores the rev_seq values of new items from the active vbucket.
MB-11037 High replication latency is observed due to the racing in pausing and resuming of the TAP connection notifier.
MB-11411 Warmup with an access log always sets the loaded document's rev-id to 1 .
Views
MB-10921 File descriptor leak is detected in views with reduce.
MB-11187 V8 crashes on memory allocation errors and closes Erlang on some indexing loads .
Server configuration
MB-11064 The default memcached connection limit was increased to the following:
max_conns_on_port_11210: 30000
max_conns_on_port_11209: 5000

Known issues in 3.0.0

The following known issues are associated with Couchbase Server release 3.0.0:

Issue Description
DCP/rebalance
MB-12226 Rebalance operation hung during the online upgrade.
MB-12125 Rebalance swap regression of 39.3% when compared with release 2.5.1.
MB-12124 Rebalance after failover, 3 to 4 nodes regression of 54.9% when compared with 2.5.1.
MB-11642 Intra-replication falling far behind under moderate to heavy workload when XDCR is enabled on the small-scale hardware (for example, AWS instances).
Views
MB-11917 One node is slow, possibly due to the Erlang scheduler.
MB-11589 Sliding endseqno during initial index build or DCP reading from disk snapshot results in longer stale=false query latency and index start-up time.
DOC-18 Due to a limitation in JavaScript, the document editor overwrites precision of long numbers.