Upgrade from LTS 2017 following Fast Tracks

From LTS 2017 to 10.1

Installation and Configuration

New Parameters

Code Changes

Replace json-lib with jackson

You should now use com.fasterxml.jackson.core:jackson-*, instead of net.sf.json-lib:json-lib.

See NXP-24093.

CSRF Protection

CSRF protection is activated by default and based on the CORS configuration, along with its allowOrigin and supportedMethods parameters, which by default doesn't allow any cross origin.

See NXP-24331.

Video Conversion Listener

The video info (duration, format, etc.) is now computed by an asynchronous work to avoid loading the blob and running ffmpeg-info synchronously. This work, schedules two asynchronous works to process the video storyboard and conversions.

Class removed: VideoAutomaticConversionListener, VideoStoryboardListener and org.nuxeo.ecm.platform.video.VideoConstants#VIDEO_CHANGED_EVENT (videoChanged).

Class added: VideoInfoWork and VideoStoryboardWork.

See NXP-24316.

Code Behavior Changes

• ecm:versionVersionableId is now indexed by Elasticsearch.
  See NXP-24114.

• Exception stack trace is no longer written in the response by default. Use parameter org.nuxeo.rest.stack.enable to enable it.
  See NXP-23861.

Deprecated APIs

@LocalDeploy

@LocalDeploy is now deprecated it must not be used anymore, use @Deploy instead that now allows you to deploy local contributions.

See NXP-22544.

CoreSession#close

CoreSession.close() is deprecated and should not be used anymore.

See NXP-24089.

Complementary Information

• Upgrade notes for 10.1
• Release notes for 10.1

From 10.1 to 10.2

Installation and Configuration

Requirements

• Elasticsearch >= 6.x is required.
• Kafka >= 1.1.0 is required.

Data

Elasticsearch

Elasticsearch 6.x is required since Nuxeo 10.2. Follow those necessary steps to upgrade:

1. Upgrade Elasticsearch version
2. Update your custom settings and mapping to follow new Elasticsearch rules
3. Re-index the repository to apply the new settings and mapping
4. Update your custom code that queries Elasticsearch directly

Upgrade Elasticsearch Version

If your indices have been created with LTS 2017 they are in Elasticsearch 5.6 format and can be read by Elasticsearch 6.x. in this case follow the full cluster restart upgrade procedure.

If your indices have been created with LTS 2016 (Elasticsearch 2.x format) or LTS 2015 (1.x format) Elasticsearch 6.x will not start, in this case an index need to be migrated to the new Elasticsearch 6.x format:

• The repository index named nuxeo by default doesn't need this migration because the repository will be re-indexed in the next step, so once this index has been backed up you can delete it.
• The sequence index named nuxeo-uidgen cannot be migrated because the _source field is disabled, Nuxeo will take care to re-create this index at startup, so once this index has been backed up you can delete it.
• The audit index named nuxeo-audit need to be migrated. Follow the reindex upgrade procedure.

Please refer to Elasticsearch documentation for more information on upgrading your Elasticsearch cluster.

Update Your Custom Elasticsearch Settings and Mapping

If you have a custom configuration for Elasticsearch it needs to be adapted to follow the new 6.x rules.

Here are the necessary changes to apply on your existing Nuxeo LTS 2017/Elasticsearch 5.6 configuration. (If you upgrade from an older configuration refer to the upgrade steps of LTS 2017).

• The default settings is located in templates/common-base/nxserver/config/elasticsearch-doc-settings.json.nxftl.
  • The ngram tokenizer will complain if the difference of max_gram and min_gram is greater than one. Either increase the max_ngram_diff, either reduce the max_gram. Note that the default Nuxeo settings defines an ngram analyzer but it is not used by default.
• The default mapping is located in templates/common-base/nxserver/config/elasticsearch-doc-mapping.json.
  • _all is a deprecated field, Nuxeo uses an explicit all_field for this purpose, the line "_all": {"enabled": false} must be removed.
  • Only one Elasticsearch type is supported per index, the line "mapping.single_type": true must be removed.

Re-Index the Repository

If your indices have been created with LTS 2017 (Elasticsearch 5.6 format) you don't need to run this step.

If you want to apply your new custom settings or mapping you have to re-index the entire repository.

When using the JSF UI this can be done from the Admin > Elasticsearch > Admin page.

Or this can be done via REST:

curl -v -X POST 'http://localhost:8080/nuxeo/site/automation/Elasticsearch.Index' -u Administrator:Administrator -H 'content-type: application/json+nxrequest' -d '{"params":{},"context":{}}'

Update Your Custom Code That Query Elasticsearch Directly

Any custom native queries done using the passthrough or code need to be reviewed to follow Elasticsearch 6 query format, for instance:

• query on boolean term must be explicitly use true or false no more 1 or 0
• query must not use the _type field.

Code Changes

Automation Scripting

The import constraints have been relaxed, it's now possible to allow specific Java classes to be used via Automation Scripting, by default we add:

• java.util.ArrayList
• java.util.Arrays
• java.util.Collections
• java.util.UUID
• org.nuxeo.runtime.transaction.TransactionHelper
• org.nuxeo.ecm.core.api.Blobs
• org.nuxeo.ecm.core.api.impl.blob.StringBlob
• org.nuxeo.ecm.core.api.impl.blob.JSONBlob
• org.nuxeo.ecm.core.api.NuxeoException

See NXP-25020 and NXP-25211.

TrashService

API Changes

The new TrashService fires dedicated events documentTrashed and documentUntrashed (hold by TrashService interface) instead of lifecycle_transition_event. The document model passed in the event won't be saved by Nuxeo at the end.

Trash Migration

As TrashService now leverages the system property ecm:isTrashed by default, you need to migrate your instance.

1. You need to replace all occurrences of ecm:currentLifeCycleState with deleted state by ecm:isTrashed in your NXQL/Page Provider/Content View, etc.
   For instance ecm:currentLifeCycleState = 'deleted' is to replace by ecm:isTrashed = 1.
2. Add the contribution from section Keeping old trash implementation.
3. In JSF UI, go to Admin > System Information > Migration, click the button next to Migration of in the trash storage model field and wait until the migration is completed.
4. Then perform an Elasticsearch re-indexation of all repository. In JSF UI, go to Admin > Elasticsearch > Admin, click the button Re-index repository and wait until the re-indexation is completed.
5. Remove the contribution added at step 2.
6. You now need to remove deleted lifecycle state from your lifecycle policies as it is deprecated and not used anymore.

Keeping Old Trash Implementation

The trash implementation has changed in 10.2. If you want to keep previous implementation relying on lifecycle state, add the following contribution:

  <require>org.nuxeo.ecm.core.trash.service.migrator</require>
  <extension target="org.nuxeo.runtime.migration.MigrationService" point="configuration">

    <migration id="trash-storage">
      <defaultState>lifecycle</defaultState>
    </migration>

  </extension>

See NXP-24850.

Code Behavior Changes

• Some test classes have been moved under nuxeo-core-io test part. But nuxeo-core-io test module is a dependency of nuxeo-core-test (former location) to enforce backward compatibility.

  TransactionalFeature has been moved from nuxeo-core-test to nuxeo-runtime-test:
  org.nuxeo.ecm.core.test.TransactionalFeature becomes org.nuxeo.runtime.test.runner.TransactionalFeature

See NXP-25197.

• New APIs are available on KeyValueStore.
  

  • Optimized storage of Long values:

    • put(String key, Long value)
    • put(String key, Long value, long ttl)
    • getLong(String key)
    • getLongs(Collection keys)

  • Atomic increment:

    • addAndGet(String key, long delta)

  See NXP-23745.

• You can add the skipAggregates=true http header when invoking the search rest endpoint to skip any aggregation computation on a page provider query.

  See NXP-25158.

Operation changes

Since Nuxeo 10.2 and 9.10-HF03, the SuggestUserEntries operation performs a full name user search, e.g. typing "John Do" returns the user with first name "John" and last name "Doe".

See NXP-24583.

Complementary Information

• Upgrade notes for 10.2
• Release notes for 10.2

From 10.2 to 10.3

Installation and configuration

Requirements

• Kafka >= 2.0.0 is required.

New Parameters

Data

Elasticsearch

An Elasticsearch RestClient Trust Store can now be configured.

The following nuxeo.conf properties can be set to define an appropriate TLS/SSL configuration for the Elasticsearch RestClient:

• elasticsearch.restClient.truststore.path
• elasticsearch.restClient.truststore.password
• elasticsearch.restClient.truststore.type
• elasticsearch.restClient.keystore.path
• elasticsearch.restClient.keystore.password
• elasticsearch.restClient.keystore.type

The following properties are deprecated (they were misnamed and are actually referring to the trustStore, not the keyStore):

• elasticsearch.restClient.keystorePath
• elasticsearch.restClient.keystorePassword
• elasticsearch.restClient.keystoreType

If a more fine-grained configuration is needed than properties, the following extension point can be used instead:

  <require>org.nuxeo.elasticsearch.defaultConfig</require>
  <extension target="org.nuxeo.elasticsearch.ElasticSearchComponent" point="elasticSearchClient">
    <elasticSearchClient class="org.nuxeo.elasticsearch.client.ESRestClientFactory">
      ...
      <option name="trustStorePath">/path/to/cacerts.jks</option>
      <option name="trustStorePassword">changeit</option>
      <option name="trustStoreType">jks</option>
      <option name="keyStorePath">/path/to/keystore.jks</option>
      <option name="keyStorePassword">changeit</option>
      <option name="keyStoreType">jks</option>
    </elasticSearchClient>
  </extension>

See NXP-26074

MongoDB

A MongoDB Client TrustStore can be configured for in-flight encryption.

The following nuxeo.conf properties can be set to define appropriate TLS/SSL configuration for MongoDB:

• nuxeo.mongodb.ssl=true
• nuxeo.mongodb.truststore.path
• nuxeo.mongodb.truststore.password
• nuxeo.mongodb.truststore.type
• nuxeo.mongodb.keystore.path
• nuxeo.mongodb.keystore.password
• nuxeo.mongodb.keystore.type

If a more fine-grained configuration is needed than properties, the following extension point can be used instead:

  <require>org.nuxeo.mongodb.connection</require>
  <extension target="org.nuxeo.runtime.mongodb.MongoDBComponent" point="connection">
    <connection id="default">
      ...
      <ssl>true</ssl>
      <trustStorePath>/path/to/cacerts.jks</trustStorePath>
      <trustStorePassword>changeit</trustStorePassword>
      <trustStoreType>jks</trustStoreType>
      <keyStorePath>/path/to/keystore.jks</keyStorePath>
      <keyStorePassword>changeit</keyStorePassword>
      <keyStoreType>jks</keyStoreType>
    </connection>
  </extension>

See NXP-26072

Redis

The following nuxeo.conf properties can be set to define appropriate TLS/SSL configuration for Redis:

• nuxeo.redis.ssl=true
• nuxeo.redis.truststore.path
• nuxeo.redis.truststore.password
• nuxeo.redis.truststore.type
• nuxeo.redis.keystore.path
• nuxeo.redis.keystore.password
• nuxeo.redis.keystore.type

If a more fine-grained configuration is needed than properties, the following extension point can be used instead:

  <require>org.nuxeo.ecm.core.redis.config</require>
  <extension target="org.nuxeo.ecm.core.redis" point="configuration">
    <server>
      ...
      <ssl>true</ssl>
      <trustStorePath>/path/to/cacerts.jks</trustStorePath>
      <trustStorePassword>changeit</trustStorePassword>
      <trustStoreType>jks</trustStoreType>
      <keyStorePath>/path/to/keystore.jks</keyStorePath>
      <keyStorePassword>changeit</keyStorePassword>
      <keyStoreType>jks</keyStoreType>
    </server>
  </extension>

See NXP-26073

Code Changes

CSRF Protection

Activation

To activate CSRF Token verification, use the following configuration:

  <extension target="org.nuxeo.runtime.ConfigurationService" point="configuration">
    <property name="nuxeo.csrf.token.enabled">true</property>
  </extension>

When this is activated, all clients accessing Nuxeo will need to get a token and provide it on all requests that are not GET/HEAD.

Note that even when the CSRF Token is not activated, other CSRF checks not using a token are still being done (using the Origin/Referer headers).

Getting the token initially

The client must use the following request with the header CSRF-Token: fetch:

GET /nuxeo
CSRF-Token: fetch

The response will contain the token in the header:

200 OK
CSRF-Token: uNTIwv3oEImb3singqJKSuJDNjM9ldVOjnwtxmFh

Passing the token

Then on every request that is not a GET/HEAD (so this applies to POST/PUT/DELETE/etc.) the client must provide the same token, either in the CSRF-Token request header or in the csrf-token request parameter:

POST /nuxeo/something
CSRF-Token: uNTIwv3oEImb3singqJKSuJDNjM9ldVOjnwtxmFh

or

POST /nuxeo/something?csrf-token=uNTIwv3oEImb3singqJKSuJDNjM9ldVOjnwtxmFh

Missing, expired or invalid token

If the token is missing, expired or invalid, the client will get a 403 Forbidden error, and a CSRF-Token: invalid header will be set:

403 Forbidden
CSRF-Token: invalid

Skipping certain endpoints

Some authentication endpoints need to be available with a POST without CSRF token checks. This can be done using for example:

  <extension target="org.nuxeo.runtime.ConfigurationService" point="configuration">
    <property name="nuxeo.csrf.token.skip" list="true">/login</property>
  </extension>

See NXP-25903

CSRF attack from local filesystem is disabled by default.

On 10.3, the behavior is still disallowed but will be allowed in dev mode.

On 7.10, 8.10 and 9.10, the behavior is now again allowed by default. To prevent local filesystem files from being allowed to POST, use the following contribution:

   <require>org.nuxeo.ecm.platform.ui.web.cors</require>
   <extension target="org.nuxeo.runtime.ConfigurationService" point="configuration">
     <property name="nuxeo.cors.allowNullOrigin">false</property>
   </extension>

See NXP-25680

Directories

New directory query APIs using a QueryBuilder:

• Session.query(QueryBuilder, fetchReferences)
• Session.queryIds(QueryBuilder)

See NXP-19262

Conversion

When calling a converter directly through its name, a check is done on its source MIME type to see if it can handle the input blob. A ConversionException is thrown if the converter can't handle it. A new configuration property nuxeo.convert.enforceSourceMimeTypeCheck is available to disable this behavior which defaults to true.

To disable the check, use the following contribution:

<require>org.nuxeo.ecm.core.convert.configuration</require>
<extension target="org.nuxeo.runtime.ConfigurationService" point="configuration">
  <property name="nuxeo.convert.enforceSourceMimeTypeCheck">false</property>
</extension>

See NXP-25840

Transient Username and External Permissions

A computed transient username is now always the same for a given base username, so only one token is generated for a given email when giving permission to an external user: that means if you invite the same external user on 2 documents, he won't have to log out from the first document to see the second one.

To enable back the uniqueness of the transient username computation, use the following contribution:

<require>org.nuxeo.ecm.core.api.properties</require>
<extension target="org.nuxeo.runtime.ConfigurationService" point="configuration">
  <property name="nuxeo.transient.username.unique">true</property>
</extension>

See NXP-25828

TrashService

org.nuxeo.isTrashed.from.deleteTransition introduced in 10.2 has been removed. Now, if you have migrated your Nuxeo instance to use the new dedicated isTrashed property, calls to CoreSession#followTransition with the delete/undelete transition are forwarded to TrashService without following the transition.

Also, these transitions are deprecated as we DO NOT follow them anymore. As a consequence, after migration or on a fresh instance, documents can't have deleted state anymore.

As a consequence of this backward mechanism, following these transitions on a proxy will remove them.

See NXP-25761

Keeping Old Comments

The comment implementation has changed in 10.3. If you want to keep your old comments, add the following contribution:

<require>org.nuxeo.ecm.platform.comment.manager.migrator</require>
<extension target="org.nuxeo.runtime.migration.MigrationService" point="configuration">
  <migration id="comment-storage">
    <defaultState>relation</defaultState>
  </migration>
</extension>

If you want to migrate comments to the new storage model, follow the Comment migration steps.

Comment Migration

To migrate comments to the new storage model:

1. In the Nuxeo Platform's JSF UI, go to Admin > System Information > Migration,
2. Click the Detect button and wait until Nuxeo isolated which comments to migrate,
3. Click the button Migrate comments from relations to the parent id property usage and wait until migration is completed,
4. Click the Detect button to make sure no comments are leftover.

Log4j 2

Nuxeo now uses Log4j 2 as its logging backend instead of Log4j. Two logging APIs are generally available in Nuxeo:

• Commons Logging
• Log4j 2 API

SLF4J is still available, but Log4j 2 API are preferred by reason of its lambda support.

Breaking change

• Log4jHelper.setQuiet and Log4jHelper.setDebug was removed due to its usage of Log4j API which are now impossible with Log4j 2. These APIs were replaced by Log4jHelper.setLevel. Compared to previous implementation, setLevel API doesn't change appender's threshold, it just acts on logger's levels.
• LogCaptureFeature.Filter has also changed without backward compatibility because it exposed a Log4j internal class. This has been replaced by corresponding Log4j 2 internal class.
• File rollover has changed. Files are now compressed using gzip and pattern has been generalised to NAME-YYYY-MM-dd.log.gz.
  For instance: server.log.2018-01-01 becomes server-2018-01-01.log.gz
• Watch log4j file for reload has been removed in favor of built-in logic in Log4j2. These properties are no longer read:
  • org.nuxeo.runtime.log4jwatch.disabled
  • org.nuxeo.runtime.log4jwatch.delay
    See Automatic Reconfiguration in Log4j 2 documentation to have watch file behavior.

Code behavior change

• LogCaptureFeature behavior has changed. Now log Result is cleaned between each tests. You'll have an isolated result to assert for @Before/@After/test methods.
• You may need to add the slf4j dependency if you were using it in your source code:

  <dependency>
  <groupId>org.slf4j</groupId>
  <artifactId>slf4j-api</artifactId>
  </dependency>
  

New

• LogCaptureFeature.FilterOn now accepts a logger class instead of a logger name
• You can now use Log4j 2 API in your code

See NXP-23863

Use NuxeoPrincipal Instead of Principal in Most APIs

The general upgrade rule is the Nuxeo code should not deal with a java.security.Principal anymore (except for internal authentication-related classes).

API changes

NuxeoPrincipal is now used instead of Principal in the methods and constructors for these classes:

• CoreSession
• DocumentEventContext
• EventContext
• EventContextImpl
• InlineEventContext
• OperationContext
• QueryFilter
• SQLQuery.Transformer
• SecurityPolicy
• SecurityPolicy.QueryTransformer
• SecurityService
• TrashService
• UnboundEventContext
• UserWorkspaceService
• WebContext

Removed APIs

These APIs were previously deprecated or internal and have been removed.

• CoreInstance.getInstance()
• CoreInstance.getNumberOfSessions()
• CoreInstance.getRegistrationInfos()
• CoreInstance.openCoreSession(repositoryName, Map context)
• CoreInstance.openCoreSession(repositoryName, Principal principal)
• DetachedNuxeoPrincipal
• SimplePrincipal (use UserPrincipal instead)

See NXP-25910

Thumbnail Resolution

In ThumbnailDocumentConverter, removed unnecessary constants:

THUMBNAIL_CONVERTER_NAME

THUMBNAIL_SIZE_PARAMETER_NAME // use ThumbnailConstants#THUMBNAIL_SIZE_PARAMETER_NAME instead

THUMBNAIL_DEFAULT_SIZE // use ThumbnailConstants#THUMBNAIL_DEFAULT_SIZE instead

See NXP-24717

REST API

REST API Enrichers

Enrichers can now apply to the blob type.

For instance, to get the links to open a LiveConnect blob in all applications associated to its MIME type, you can use the enrichers-blob: appLinks enricher. Any blob property, e.g. file:content, will then be enriched in the following way:

{
  "file:content": {
    "name": "...",
    "mime-type": "...",
    ...,
    "appLinks": [
      {
        "appName": "...",
        "icon": "...",
        "link": "..."
      },
      ...
    ]
  }
}

Note: since such a blob enricher applies to a BlobProperty, the backing class should extend AbstractJsonEnricher<BlobProperty> as in the example below:

public class BlobAppLinksJsonEnricher extends AbstractJsonEnricher<BlobProperty> {

    @Override
    public void write(JsonGenerator jg, BlobProperty blobProperty) throws IOException {
        ...
    }
}

See NXP-26126

REST API Endpoints

The REST API now produces and consumes only application/json as content type:

• application/json+nxentity is now never returned as content type response; the server does not expect it as content type request.
• application/json+nxrequest content type should not be used anymore. It still works but it's deprecated server side. application/json should be used instead when POSTing to automation.
• application/json+esentity which was never used was also removed.

Breaking Change:

• org.nuxeo.ecm.automation.jaxrs.io.documents.JsonESDocumentWriter, located in nuxeo-automation-io has moved to org.nuxeo.elasticsearch.io.JsonESDocumentWriter, located in nuxeo-elasticsearch-core. The JsonESDocumentWriter does not implement anymore MessageBodyWriter<DocumentModel>, you may need to update your code.

See NXP-25036

Code Behavior Changes

• To activate OAuth 1, an <authenticationChain> must be defined to include <plugin>OAUTH1_AUTH</plugin>.
  See NXP-25975

• Fulltext maximum size is now 128 KB by default. To change this, the repository configuration can be updated to use another fieldSizeLimit value:

  <fulltext ... fieldSizeLimit="1048576">
    ...
  </fulltext>
  

  A value of 0 means no limit.

  See the full text documentation for more information.

  See NXP-25716

Operation Changes

Rename operations WebUI.* related to JSF

See NXP-22232

Relocate nuxeo-platform-collections in nuxeo-service

Since Nuxeo Platform 10.3, Collection Core types (Collection, Favorites, etc.), schema and facets have been merged in CoreExtensions.xml. As a result:

• nuxeo-platform-collections-jsf has disappeared. Resources and classes have been dispatched in nuxeo-jsf/nuxeo-platform-webapp-base and nuxeo-dm/nuxeo-platform-webapp-types
• All Collection-related automation operations have been moved to nuxeo-automation-features
• Collection and Favorites services have been kept but moved under nuxeo-services/nuxeo-collections
• UserWorkspaces service now contributes where should be located User's collections and Favorites by implementing CollectionLocationService.java
• You can use CollectionFeature.java to test a feature relying on Collections.
  You need to add the following pom dependency:

  <dependency>
   <groupId>org.nuxeo.ecm.platform</groupId>
   <artifactId>nuxeo-platform-collections-core</artifactId>
   <type>test-jar</type>
   <scope>test</scope>
  </dependency>
  

See NXP-25543

Image Recompute

Added the Picture.RecomputeViews operation for administrators.

See NXP-25791

Addons

Nuxeo Drive - Server APIs Changes

Added

• Added to FileImporter:

  boolean isOneToMany();
  

  to flag an importer as "one-to-many", ie. it creates more than one document for the given blob, typically CSVZipImporter or ExportedZipImporter.

See NXP-25797

Deprecated

• The following operations have been deprecated:

  GetRepositoriesOperation
  NuxeoDriveAddToLocallyEditedCollection
  NuxeoDriveCanMove
  NuxeoDriveGenerateConflictedItemName
  NuxeoDriveGetClientUpdateInfo
  NuxeoDriveWaitForAsyncCompletion
  

• The following operation parameters has been deprecated:

  NuxeoDriveGetChangeSummary#lastSyncActiveRootDefinitions
  

See NXP-24885

Remove Usage of Deprecated Code in Nuxeo-Drive-Server

Removed as deprecated since 7.10 or less:

FileSystemItemFactory#getDocumentByFileSystemId(String id, Principal principal)

FileSystemItemManager#getSession(String repositoryName, Principal principal)

NuxeoDriveManager#setChangeFinder(FileSystemChangeFinder changeFinder)

NuxeoDriveActions#UPDATE_SITE_URL_PROP_KEY
NuxeoDriveActions#SERVER_VERSION_PROP_KEY

NuxeoDriveCreateFile#name

NuxeoDriveGetChangeSummary#lastSyncDate

NuxeoDriveGetTopLevelChildren

Removed as should have been deprecated since 7.10 or less, see NXP-14826:

FileSystemChangeFinder#getFileSystemChangesIntegerBounds(CoreSession session, Set<IdRef> lastActiveRootRefs, SynchronizationRoots activeRoots, Set<String> collectionSyncRootMemberIds, long lowerBound, long upperBound, int limit)

FileSystemChangeFinder#getCurrentDate();

NuxeoDriveManager#getChangeSummaryIntegerBounds(Principal principal, Map<String, Set<IdRef>> lastSyncRootRefs, long lowerBound)

and changed

FileSystemChangeFinder#getFileSystemChanges(CoreSession session, Set<IdRef> lastActiveRootRefs, SynchronizationRoots activeRoots, long lastSuccessfulSyncDate, long syncDate, int limit)

NuxeoDriveManager#getChangeSummary(Principal principal, Map<String, Set<IdRef>> lastSyncRootRefs, long lastSuccessfulSync)

to

FileSystemChangeFinder#getFileSystemChanges(CoreSession session, Set<IdRef> lastActiveRootRefs, SynchronizationRoots activeRoots, Set<String> collectionSyncRootMemberIds, long lowerBound, long upperBound, int limit)

NuxeoDriveManager#getChangeSummary(Principal principal, Map<String, Set<IdRef>> lastSyncRootRefs, long lowerBound)

See NXP-25844

Nuxeo CSV

Dates in CSV files must be formatted using the W3C format. More information on the W3C documentation.

It is possible to use the old legacy date format (without time information) by setting the configuration property nuxeo.csv.import.legacyDateFormat to true:

<require>org.nuxeo.ecm.csv.core.properties</require>
<extension point="configuration" target="org.nuxeo.runtime.ConfigurationService">
  <property name="nuxeo.csv.import.legacyDateFormat">true</property>
</extension>

See NXP-25219

Nuxeo Diff

Deprecated ContentDiffHelper#getHtmlConversionBlackListedMimeTypes.

Added:

org.nuxeo.ecm.diff.content.MimeTypesDescriptor
org.nuxeo.ecm.diff.content.MimeTypeDescriptor
ContentDiffAdapterManager#getHtmlConversionBlacklistedMimeTypes()

Added the htmlConversionBlacklistedMimeTypes extension point to the org.nuxeo.ecm.diff.content.adapter.ContentDiffAdapterManagerComponent component:

  <extension-point
    name="htmlConversionBlacklistedMimeTypes">
    <documentation>
      @since 10.10

      Allows to contribute the list of blacklisted mime types for HTML
      conversion.

      By default, contributing a list merges its mime types
      with the existing ones.
      To remove a mime type, use enabled=false.
      <code>
        <extension
          target="org.nuxeo.ecm.diff.content.adapter.ContentDiffAdapterManagerComponent"
          point="htmlConversionBlacklistedMimeTypes">
          <mimeTypes>
            <mimeType>application/msword</mimeType>
            <mimeType>application/rtf</mimeType>
            <mimeType enabled="false">application/pdf</mimeType>
          </mimeTypes>
        </extension>
      </code>

      To override the whole list, use override="true".
      <code>
        <extension
          target="org.nuxeo.ecm.diff.content.adapter.ContentDiffAdapterManagerComponent"
          point="htmlConversionBlacklistedMimeTypes">
          <mimeTypes override="true">
            <mimeType>application/msword</mimeType>
          </mimeTypes>
        </extension>
      </code>
    </documentation>
    <object
      class="org.nuxeo.ecm.diff.content.MimeTypesDescriptor" />
  </extension-point>

See NXP-25208

Complementary Information

• Upgrade notes for 10.3
• Release notes for 10.3

From 10.3 to LTS 2019

Data

Binaries

In addition to the default binary store used for the document repository, Nuxeo now uses additional binary stores to store transient binaries. With the standard configuration that stores the default binaries under the directory defined by repository.binary.store, you should take into account the fact that these new binary stores will be created as siblings of this configured directory. For instance if you have repository.binary.store=/my/binaries/path then additional binary stores will be created at /my/binaries/path_transient_xyz for various values of xyz.

Therefore, if you require a mount point (like a Docker volume) for all the binaries in order to share them in a cluster configuration, we recommend to mount for instance /var/lib/nuxeo/binaries and define repository.binary.store=/var/lib/nuxeo/binaries/binaries, so that additional binary stores like /var/lib/nuxeo/binaries/binaries_transient_xyz will still be under the mount point.

This does not apply to binaries stored in S3 or in other locations that are not on the filesystem.

See NXP-26594

Redis Activation

Activating Redis is now done through a template. Previously, it was enough to do:

nuxeo.redis.enabled=true

But now a redis template must be added instead:

nuxeo.templates=default,...,redis

See NXP-26553

Code Changes

Bulk Action Framework Contextual Information

The way we store contextual information for the Bulk Action Framework (BAF) has changed. Instead of having BULK_ID:status, we have status:BULK_ID (for instance).

This makes all commands submitted before upgrade not compatible with the Bulk Action Framework after the upgrade.

There's no impact on finished command except that status can't be retrieved after the upgrade, but acceptable as it's transient data.

There's an impact on running command because after upgrading to Nuxeo Platform LTS 2019 (10.10), BAF won't be able to retrieve contextual information to finish submitted command before upgrade. As it, we advise to upgrade your Nuxeo Platform from 10.3 to 10.10 after all BAF processing has finished.

Otherwise, a manual intervention on the KeyValueStore is possible to reverse these key/value:

• BULK_ID:status by status:BULK_ID
• BULK_ID:command by command:BULK_ID

See NXP-26559

Rendition

• Removed as deprecated since 7.2:

  RenditionServiceImpl#automationService
  
  RenditionServiceImpl#getAutomationService()
  

• Deprecated:

  RenditionServiceImpl#getRenditionDefinition(String name)
  
  RenditionServiceImpl#storeRendition(DocumentModel sourceDocument, Rendition rendition)
  

• Added:

  RenditionService#getAvailableRenditionDefinition(DocumentModel doc, String renditionName)
  
  RenditionServiceImpl#storeRendition(DocumentModel sourceDocument, Rendition rendition, RenditionDefinition renditionDefinition)
  

See NXP-25501

KeyValueStore Implementation in SQL

A new Key/Value Store based on SQL is available. To configure a server to use it, use:

<extension target="org.nuxeo.runtime.kv.KeyValueService" point="configuration">
  <store name="default" class="org.nuxeo.ecm.core.storage.sql.kv.SQLKeyValueStore">
    <property name="datasource">jdbc/nuxeo</property>
    <property name="table">kv</property>
  </store>
</extension>

See NXP-25604

SAML Icon Displayed on Login Page

• Added:

  LoginScreenHelper#registerSingleProviderLoginScreenConfig(String name, String iconUrl, String link, String label, String description, LoginProviderLinkComputer computer)
  
  LoginScreenHelper#unregisterLoginScreenConfig(LoginScreenConfig config)
  
  LoginProviderLink(String name, String iconPath, String link, String label, String description, LoginProviderLinkComputer urlComputer)
  
  LoginScreenConfig(LoginProviderLink provider)
  
  PluggableAuthenticationService#registerLoginScreenConfig(LoginScreenConfig config)
  
  PluggableAuthenticationService#unregisterLoginScreenConfig(LoginScreenConfig config)
  

• Deprecated:

  LoginScreenHelper#registerLoginProvider(String name, String iconUrl, String link, String label, String description, LoginProviderLinkComputer computer)
  
  LoginScreenConfig#registerLoginProvider(String name, String iconUrl, String link, String label, String description, LoginProviderLinkComputer computer)
  

See NXP-25837

FileManager

All FileManager#createDocumentFromBlob methods have been deprecated, you should now use the FileManager#createOrUpdateDocument(FileImporterContext) method to create or update documents from blobs.

FileImporter#create method has been deprecated, you should now use the FileImporter#createOrUpdate(FileImporterContext) method.

The FileImporterContext class contains everything needed to create a document from a blob.

Before:

DocumentModel doc = fileManager.createDocumentFromBlob(session, blob, parentPath, /* overwrite */ true, /* filename */ "foo.html");

After:

FileImporterContext context = FileImporterContext.builder(session, blob, parentPath)
    .overwrite(true)
    .fileName("foo.html")
    .build();
DocumentModel doc = fileManager.createOrUpdateDocument(context);

Use FileImporterContext.Builder#excludeOneToMany to allow excluding "one-to-many" importers (ie. the ones creating more than one document for the given blob, typically CSVZipImporter or ExportedZipImporter) when selecting the importer. This is used by Drive to not try to explode a CSV import ZIP for instance.

The FileImporterContext#getBlob's file name is used by default now, but you can still override it with FileImporterContext.Builder#fileName(String).

If the underlying FileImporter handles it, there is now a way to let the FileManager create a document from a blob without persisting it using FileImporterContext.Builder#persistDocument(boolean). It's then the caller's responsibility to actually persist the document.

FileImporterContext fileCreationContext = FileImporterContext.builder(session, blob, parentPath)
    .persistDocument(false)
    .build();
DocumentModel doc = fileManager.createOrUpdateDocument(fileCreationContext);
// doc is not persisted
// doc.isDirty() => true
// doc.getId() => null
// do domething on doc
doc = session.createDocument(doc);
// doc is persisted

Breaking change

Custom importers implementing FileImporter or extending AbstractFileImport must be updated to override the #createOrUpdate(FileImporterContext)} method instead of the #create method. Not overriding the #createOrUpdate(FileImporterContext) method will make your custom importer useless as it will never be called.

See this commit for an example of migration to #createOrUpdate(FileImporterContext).

See NXP-26178

VersioningService

org.nuxeo.ecm.core.versioning.VersioningService has been deprecated, you should now use org.nuxeo.ecm.core.api.versioning.VersioningService.

See NXP-26423

Operation Changes

Regenerate Thumbnails on Demand

There is a new operation called RecomputeThumbnails operation, available for administrators to let them regenerate thumbnails on demand.

Sample call:

curl -v -H 'Content-Type:application/json' -d '{"params": {"query": "SELECT * FROM Document WHERE ecm:mixinType = \"Thumbnail\" AND thumb:thumbnail/data IS NULL AND ecm:isVersion = 0 AND ecm:isProxy = 0 AND ecm:isTrashed = 0"}, "context": {}}' -u
Administrator:Administrator
http://localhost:8080/nuxeo/site/automation/RecomputeThumbnails

See NXP-26282

Create Document From Blob with Properties

A new atomic operation FileManager.ImportWithProperties has been added allowing to create a document from a blob and setting properties at the same time.

See FileManagerImportWithProperties

See NXP-26178

Complementary Information

• Upgrade notes for LTS 2019
• Release notes for LTS 2019