As of June 2021, with the release of Neo4j 4.3.0 halin is now deprecated.
The software will continue to be available as a GraphApp and via https://halin.graphapp.io/; and if you currently depend on it, it will not break or disappear.
Halin will not support all 4.3.0 features, and you may encounter incompatibilities when using Halin with this version. Halin will continue to work with 3.5 <= Neo4j <= 4.2 as it always did.
New features & changes
- Upgrades to work with the Neo4j 4.2 series
- Removed support for Neo4j 3.4.x; newer drivers are incompatible with older bolt protocol
- Neo4j Aura is not supported in any version or tier
- New connection dialog per member, allowing individual killing of connections
Bug fixes and maintenance
- Upgraded dependencies, addressing security bulletins, including Neo4j driver 4.2.1
- Fixed unmounted component state change issues
- Fixed pagination on very small result sets
- Fixed #157 running Halin in kubernetes
- Connection details may be passed by URL string, as in:
http://halin-host.com/?host=mydb&port=7687
- Fixed minor Desktop deployment issue
- Minor dependency updates for security purposes
- Disable Sentry error reporting in most cases.
- Halin now shows committed heap in the overview, not configured heap
- Improved version parsing for the various kinds of neo4j encountered
- Cluster timeseries charts can be paused when all underlying data feeds are in an error state
- Halin does not support Neo4j 4 on Aura
- Halin can now connect to and monitor clusters with (partially) broken routing tables, which can occur when a member is falling offline, or when one member (i.e. a RR) has an invalid advertised address.
- Built-in system roles (expanded to include PUBLIC in Neo4j 4.1) cannot be removed, and their privileges cannot be revoked.
- Test infrastructure now tests against Neo4j Enterprise 4.1.0
- Better error handling for common Neo4j errors
- DB-specific data feeds get stopped & started along with the underlying database stop/restart.
- Fixed some data feed errors caused by underlying race conditions in data return
- Revamped garbage collection to be a scatterplot of instantaneous GC
- Moved Transactions view to the database pane, tailored per-database for 4.0 support.
- Removed File Descriptor view, as less often requested/problematic.
- Deprecated some components
- Refactored the event API for core Metrics
- A bug in the generation of privilege commands was fixed, which could give the wrong results
when the selected graph was
* - A "copy role" option was added for Neo4j 4.0 installs; the "copy" icon button in Actions on the role administration page
- Added alerts for certain memory conditions, such as heap re-allocation and imminent out-of-memory errors (OOMs)
- Fixes #132: Halin will warn you when you are approaching the need for high_limit storage.
- Members with non-standard ports will now have their port displayed in their label. This helps separate members running on the same host.
- Main nav menu is labeled for help with navigation
- Thanks to Hugo for several points of feedback on this release.
- Fixes #127: Neo4j 3.5 standalone members appear properly, not as LEADER
- Fixes #126: Metric chart improvements for readability & precision
- Fixes #128: excessive error reporting
- Upgrades to sentry dependencies
- Simplify the data feed statistics page into a table
- Fixes #121 identify halin with global user-agent string in bolt driver
- Fixes #122 confusion between open and available OS file descriptors
- Fixes #97 omit 'parameters' from query/task view for performance reasons
- Fixes #125 database names may contain dots and dashes
- React scripts updates, which addresses numerous github security notices
- Change base docker image to upgrade node, yarn, and react-scripts
- Status for both all cluster members and all databases. Users can independently see the mapping between the two
- Halin API exposes a better cluster-wide API for fine-grained cluster control, including ClusterMemberSet, DatabaseSet, and feature probes.
- Full support for GRANT/REVOKE/DENY of fine-grained privileges under Neo4j 4.0
- Support for tracking LEADER/FOLLOWER status in multi-database arrangements of 4.0
- Halin knows how to detect presence of Fabric enabled databases
- Removed Neo4j Desktop API dependencies, permitting upgrades to key components (semantic-ui-react). As a result, users must always log in even when using as a graph app.
- Initial support for Neo4j 4.0 as of MR03; multiple databases now supported!
- New database pane that allows creating, deleting, stopping and starting databases.
- Fixes #105, warn on default tx_log retention policy
- Because Neo4j 4.0 is not yet final, this is a beta release subject to change. Support for Neo4j 3.4 and 3.5 remains steady, but Neo4j 4.0 support is provisional only.
- Fixes #96, adds labels to GC Pause times
- Addresses #98, more help for troubleshooting Halin connection issues
- Addresses #99, which detects non-standard cluster routing TTL in the diagnostics rules
- Addresses #100, when Neo4j desktop doesn't pass the appropriate token to connect
- Merges PR #94 (thanks to @wagnerjt) which improves various components to, among other things, use pure functional components where possible.
- Fixed packaging error that affected some Neo4j Desktop users.
- Added contributions file (closes #95)
- Reorganize task pane to better display all transactions, even those with no currently executing query
- Allow killing transactions in the Task pane (Neo4j >= 3.5)
- Fixed minor packaging issue reported by community user
- New UI layout which is more expand-able for larger clusters.
- New metrics feature for Neo4j instances with most recent APOC installed; view/track server-side metrics.
- Refactored component structure for better maintainability
- Added connection concurrency controls which have the effect of improving startup time over high-latency networks, e.g. when Neo4j is network distant from the Halin client.
- Various improvements for feature filtering for read-only users of Halin
- Thanks to D. Shiposh, D. Fauth, R. Kharwar, and others who provided valuable feedback on security & usability fixes.
- Fixes and/or closes #90, #86, #60, #30, #88, #69
- Enabled CSV download of query stats, and active tasks (#78)
- Log view screen allows inspecting query.log if it exists (#77)
- Enabled log parsing for some logs, downloadable as CSV (#77)
- Permitted downloading certain tables (configuration, plugins, others) as CSV
- Several improvements to the UI for managing users/roles (#82)
- Support for auth methods used by Neo4j Cloud EAP (#72)
- Put linting in place codebase wide, integrated into CI build (#83)
- Expanded use of explainers, with more content for users.
- Logs pane added to individual nodes, so that you can see the content of debug.log, neo4j.log, and security.log, if you are using a recent APOC distribution
- "Active Queries" component has been replaced by "Tasks" panel (Machine->Performance) which adds the ability to inspect transaction metadata
- Expanded unit testing
- Bug fixes for address handling, Neo4j int handling
- Fixes #71 member address does not change when role updates
- Fixes #73 too few arguments to std
- Partially addressing #60, log & metric access.
- Minor improvements to the settings pane
- Fixed somewhat exotic bug that could cause problems removing roles from users under certain versions of Neo4j
- Introduced a number of unit tests & testing approach
- Documented almost all queries Halin executes
- Implemented session pooling to prevent extra network round-trips to Neo4j, which reduces overall latency and improves performance.
- Centralized querying in ClusterNode#run permitting the use of central TX metadata for things that Halin sends to the server. #65
- Adds query performance profiling ("Query Performance" tab under each machine) which permits temporary sampling of live running queries, and performance analysis.
- Adds "Explainers", info icons to most sections, that describe what the data means and provides links to appropriate Neo4j documentation
- Greatly improves compatibility checking for various features of Neo4j.
- #59 Makes the leader in a cluster more visually distinct in a timeseries graph
- #57 prevents connection errors by better validating inputs when connecting
- #62 fixes diagnostic collection for non-native auth (i.e. LDAP)
- Defaults roles to [] for community edition, which does not support roles.
- Fixes an issue identified where incorrect roles were reported
- Fixes #55; removes DBcount component and institutes a new plugins / CypherSurface component to consolidate view of functions and procedures
- Fixes #54; adds an "Other" column to store file size tracking, to indicate when users have extraneous files in their graph.db folders that are taking up space. Thanks to @dfgitn4j for the suggestion.
- (Optional opt-in) you may report diagnostic information to help improve Halin & Neo4j
- Centralize driver error handling
- Fixes #51, #48 - Halin errors on dbms.security.auth_enabled=false
- Fixes #49 - Halin errors in some cases when a non-admin user is used for login
- Fixes data display issues in the store file sizes component to make them human readable
- Fixes bug where address of machine involved in role changes did not update properly
- Halin can now gather diagnostics and render advice about Neo4j Community
- The user management pane now works for Neo4j Community (no longer just enterprise)
- Improvements to the diagnostic data format, and better tracking of node performance.
- Bugfixes: #40, #41, #42, #44
- Hotfix release for deploy errors
- Bugfixes to support Neo4j 3.5.0 (#38) community & enterprise.
- Fixes browser compatibility issues for older browsers (#36)
- Via
npm run gatherit is now possible to gather diagnostics non-interactively. - Fixed page cache advisor bug where certain memory values were mis-identified.
- Improved diagnostic output to include kernel version number and native auth support information for each cluster node
- Added diagnostic rule to detect version consistency amongst machines in a cluster.
- Added the basics of a CircleCI build system
- Added Docker support. Halin can be built and run inside of docker.
- Fixed Neo4j Desktop bug that occurred when no database was active (#31)
- Added several suggestions from Dave Shiposh; heap size checking, page cache size checking (#29, #28)
- Disabled User Management tab when Neo4j cluster does not support native authentication (#27)
- Improved freshness calculations for degrading connections
- Improved connection troubleshooting
- Key dependency updates
- Certain cluster lifecycle events (like leader re-elections) can be detected and seen in the diagnostic tab
- Status icons (green/yellow/red) on the top tab bar showing responsiveness of each cluster node
- Better display/disabling for Neo4j Community, when components cannot be rendered. Some functions do not apply to enterprise.
- Documentation and license on repo.
- Support for Neo4j Community! Albeit with some components disabled whose data is not available outside of Enterprise
- Selectively disable certain components when user lacks the 'admin' role, rather than breaking badly
- Sentry integration for tracking errors encountered
- Fixed edge case bugs where certain kinds of JMX queries do not return any records in Neo4j Community
- Introduced the concept of a cluster timeseries, showing one data element for all nodes in the cluster.
- Based on that, added a Cluster Overview tab with cluster timeseries for heap size, GC pause, page cache faults, transactions open, used memory, file descriptors, and others, following templated Grafana dashboards suggested by DGordon.
- Introduced the concept in Diagnostics of a "Configuration Diff Tool" allowing users to quickly find where configuration disagrees across all nodes of the cluster.
- Data feed stats on the settings page
Bugfixes
- Issue #14: systems which don't have swap memory show an eternal loading spinner
- Issue #13: time windows are inappropriately reset when switching between tabs
- Fix issue #10; 127.0.0.1 backups are not recognized as local
- Improved timeseries moving window behavior
- Added an OS sub-window with operating system memory statistics
- New disk usage timeseries monitor
- Bugfix: you can't escape the first connection modal.
- Halin can now run independent of Neo4j Desktop as a web app. (
yarn start) - Scripts to deploy to AWS S3
- Whether or not cluster connections are encrypted depends on Desktop settings and user preference via the connection dialog
Lots and lots of GraphConnect 2018 Feedback
- Promoted User Management to cluster-wide operations
- Added links to configuration entries reference documentation
- Removed passwords and sensitive information from diagnostic dumps
- Added "Ping" component to diagnostic page to measure round-trip times
- Changed download diagnostic package button to be more clear
- Initialize cluster connections on startup for better performance
- Advisor results come broken out by machine
- Added checking for backup status, network security to advisor
- Added app footer with link to the page
- Fixed several issues with package.json
- Corrected icon path
- Icons and package.json formatting to make Halin play
- nicely with the Neo4j Desktop UI
- first working published version