Humio Server 1.6.3 Archive (2019-09-25)

Version?Type?Release Date?Availability?End of Support

Security

Updates

Upgrades

From?

Config.

Changes?
1.6.3Archive2019-09-25

Cloud

2020-11-30No1.5.19No

Available for download two days after release.

Hide file hashes

Show file hashes

Dashboard parameters improvements and Bug Fixes. Data migration is required: Hash filters need rebuilding.

Dashboard parameters can depend on each other. Fixed various small UI bugs in data table. Improvements to event list column headers.

New features and improvements

  • Other

    • File based parameters on dashboards can now filter parts of a file out, by specifying a subset of entries in the file that should be used. This filtering can also be based on other parameters, so entries pulled from the file can depend on e.g. a query based parameter.

Fixed in this release

  • Functions

    • Using dropEvent() in a parser did not work when using the "Run Tests" button.

  • Other

    • EventList column header menu opens on click now, instead of on mouse hover.

    • Exporting a search (Or using /query endpoint in other contexts) would fail if any node was down even when the files needed to satisfy the search were available on other nodes. Note that a search in progress will still fail if a node goes missing while the search runs. (Searches in the UI restart in this case but that is not possible for an export.)

    • MAX_EVENT_SIZE defaults to 1 MB. Increasing this may have adverse affects for overall system performance.

    • When setting up Humio the server will refuse to start if Kafka is not ready in the sense that the number of live Kafka brokers is less than number of Kafka bootstrap hosts in the configuration for Humio.

    • Regex matching gets rejected at runtime if it spends too many resources.

    • Improved names and states in thread dumps and added a group field to the traces. Run #type=humio class=threaddump state=RUNNABLE | timechart(group,limit=50,span=10s) in the Humio repo to get an idea of variations in what the CPU time is being spent on.

    • The Show in context window on event list would "jump" when used and a live query on dashboards.

    • Fix issue that made the timestamp column wrap on some platforms.

    • In Chrome, it was sometimes not possible to rename a dashboard, clone a dashboard, duplicate a widget, and other actions. This has been fixed.

    • LDAP login code rewritten.

    • Make JSON word-wrapping work when a column is syntax highlighted.

    • Fix issue with layout of pagination of table widgets in dashboards overflowing when it has a horizontal scroll bar

    • Latin-1 characters (those with code point 128 - 255) were not added correctly to hash filters. To fix this, Humio needs to rebuild the existing hash filters: The old hash files get deleted, and a new file prefix "hash5h3" is applied to the new files. This will be done in the background after updating to this version. For estimation of time to complete use a rate of .0GB/core/hour of original size. While rebuilding hash filter files the system will have a higher load from this and from searches that would benefit from the filters but need to run without them.

    • HASHFILTER_MAX_FILE_PERCENTAGE defaults to 50. Hash filter files that are larger than this relative to their segment file do not get created. This trades the work required to scan them on search for disk space for files that are not very large.

    • Replication of segment files among nodes now runs in multiple threads to allow faster restore form peers for a failed node.

    • Previously, exporting data from queries with parameters would always fail. This now works as expected.

    • MAX_JITREX_BACKTRACK default to 1.0.0: Limits CPU resources spent in a regex match, failing the search if exceeded.

    • Fix issue where streaming queries failed when a node in the cluster was unavailable.

    • The Event List widget no longer shows column menus on dashboards. Editing was not possible, but the menus would open anyway.