Falcon LogScale 1.153.4 LTS (2024-12-17)
Version? | Type? | Release Date? | Availability? | End of Support | Security Updates | Upgrades From? | Config. Changes? |
---|---|---|---|---|---|---|---|
1.153.4 | LTS | 2024-12-17 | Cloud | 2025-09-30 | Yes | 1.112 | No |
Hide file hashes
TAR Checksum | Value |
---|---|
MD5 | 89da4587a8997aaced15238e84a88c29 |
SHA1 | 54a7c0d8481492beef15c2af19217a90ee415ca9 |
SHA256 | 1219274fef386c080ab7a02410a6dc16bed88b3a49c45f072c39c6bd5d071e73 |
SHA512 | 50effa4373b048ec699ddb0a40bf29da0865723e6e05d1b446ce3ffd6c05df8085c17be09652541ec507559fa6a42b38faddf3e224ed5975c55da4363cd7a75f |
Docker Image | Included JDK | SHA256 Checksum |
---|---|---|
humio | 22 | 2bfe64dc45f561eef134c2d1d3c1c2fdd0173b474aef160aa3dcdf42a56e14a8 |
humio-core | 22 | b636b5f96d84b398a75b37aefef3162ca5b9a9213ce1834f4181a4b433801a05 |
kafka | 22 | a8c068517a4decedea274df8426b78f496e2382eca6388aa65ea63736bb70459 |
zookeeper | 22 | bad8d6ec1dfcb589bac951cb69a22169141d2fd514c0924bf3c8d31cedbd699c |
Download
These notes include entries from the following previous releases: 1.153.1, 1.153.3
Bug fixes and updates.
Breaking Changes
The following items create a breaking change in the behavior, response or operation of this release.
Removed
Items that have been removed as of this release.
Installation and Deployment
The previously deprecated
jar
distribution of LogScale (e.g.server-1.117.jar
) is no longer published starting from this version. For more information, see Falcon LogScale 1.130.0 GA (2024-03-19).The previously deprecated
humio/kafka
andhumio/zookeeper
Docker images are now removed and no longer published.API
The following previously deprecated KAFKA API endpoints have been removed:
POST
/api/v1/clusterconfig/kafka-queues/partition-assignment
GET
/api/v1/clusterconfig/kafka-queues/partition-assignment
POST
/api/v1/clusterconfig/kafka-queues/partition-assignment/set-replication-defaults
GET
/api/v1/clusterconfig/kafka-queues/partition-assignment/id
Configuration
The obsolete configuration parameters
AUTOSHARDING_TRIGGER_DELAY_MS
andAUTOSHARDING_CHECKINTERVAL_MS
have been removed due to autosharding being handled by rate monitoring and not by ingest delay anymore.Other
Unnecessary
digest-coordinator-changes
anddesired-digest-coordinator-changes
metrics have been removed. Instead, the logging in theIngestPartitionCoordinator
class has been improved, to allow monitoring of when reassignment of desired and current digesters happens — by searching forWrote changes to desired digest partitions
/Wrote changes to current digest partitions
.
Deprecation
Items that have been deprecated and may be removed in a future release.
The
server.tar.gz
release artifact has been deprecated. Users should switch to theOS/architecture-specific server-linux_x64.tar.gz
orserver-alpine_x64.tar.gz
, which include bundled JDKs. Users installing a Docker image do not need to make any changes. With this change, LogScale will no longer support bringing your own JDK, we will bundle one with releases instead.We are making this change for the following reasons:
By bundling a JDK specifically for LogScale, we can customize the JDK to contain only the functionality needed by LogScale. This is a benefit from a security perspective, and also reduces the size of release artifacts.
Bundling the JDK ensures that the JDK version in use is one we've tested with, which makes it more likely a customer install will perform similar to our own internal setups.
By bundling the JDK, we will only need to support one JDK version. This means we can take advantage of enhanced JDK features sooner, such as specific performance improvements, which benefits everyone.
The last release where
server.tar.gz artifact
is included will be 1.154.0.The lastScheduledSearch field from the
ScheduledSearch
datatype is now deprecated and planned for removal in LogScale version 1.202. The new lastExecuted and lastTriggered fields have been added to theScheduledSearch
datatype to replace lastScheduledSearch.
Behavior Changes
Scripts or environment which make use of these tools should be checked and updated for the new configuration:
Installation and Deployment
The default
cleanup.policy
for the transientChatter-events topic has been switched fromcompact
todelete,compact
. This change will not apply to existing clusters. Changing this setting todelete,compact
via Kafka's command line tools is particularly recommended iftransientChatter
is taking up excessive space on disk, whereas it is less relevant in production environments where Kafka's disks tend to be large.Automation and Alerts
Aggregate and filter alert types now both display an Error (red) status if starting the alert query times out after 1 minute.
For more information on alert statuses, see Monitoring Alerts.
Storage
Reduced the waiting time for redactEvents background jobs to complete.
The background job will not complete until all mini-segments affected by the redaction have been merged into full segments. The job was pessimistically waiting for
MAX_HOURS_SEGMENT_OPEN
(30 days) before attempting the rewrite. This has been changed to wait forFLUSH_BLOCK_SECONDS
(15 minutes) before attempting the rewrite, this means, while some mini-segments may not be rewritten for 30 days, it is uncommon. If a rewrite is attempted and encounters mini-segments, it is postponed and retried later.For more information, see Redact Events API.
Configuration
When global publish to Kafka times out from digester threads, the system would initiate a failure shutdown. Instead, from this 1.144 version the system retries the publish to Global Database indefinitely for those specific global transactions that originate in a digester thread. If retries occur, these get logged with an error executeTransactionRetryingOnTimeout: unable to execute transaction for global, retrying.
Autoshards no longer respond to ingest delay by default, and now support round-robin instead.
Functions
Prior to LogScale v1.147, the
array:length()
function accepted a value in thearray
argument that did not contain brackets[ ]
so thatarray:length("field")
would always produce the result0
(since there was no field named field). The function has now been updated to properly throw an exception if given a non-array field name in thearray
argument. Therefore, the function now requires the given array name to have[ ]
brackets, since it only works on array fields.
Upgrades
Changes that may occur or be required during an upgrade.
Installation and Deployment
The minimum version of Java compatible with LogScale is now 21. Docker users, and users installing the release artifacts that bundle the JDK, are not affected.
It is recommended to switch to the release artifacts that bundle a JDK, because LogScale no longer supports bringing your own JDK as of release 1.138, see Falcon LogScale 1.138.0 GA (2024-05-14)
The JDK has been upgraded to 23.0.1
New features and improvements
Security
When extending Retention span or size, any segments that were marked for deletion — but where the files remain in the system — are automatically resurrected. How much data you reclaim via this depends on the
backupAfterMillis
configuration on the repository.For more information, see Audit Logging.
Installation and Deployment
The Docker containers have been configured to use the following environment variable values internally:
DIRECTORY=/data/humio-data
HUMIO_AUDITLOG_DIR=/data/logs
HUMIO_DEBUGLOG_DIR=/data/logs
JVM_LOG_DIR=/data/logs
JVM_TMP_DIR=/data/humio-data/jvm-tmp
This configuration replaces the following chains of internal symlinks, which have been removed:/app/humio/humio/humio-data
to/app/humio/humio-data
/app/humio/humio-data
to/data/humio-data
/app/humio/humio/logs
/app/humio/logs
/app/humio/logs
to/data/logs
This change is intended for allowing the tool scripts in
/app/humio/humio/bin
to work correctly, as they were previously failing due to the presence of dangling symlinks when invoked via docker run if nothing was mounted at/data
.
UI Changes
LogScale administrators can now set the default timezone for their users.
For more information, see Setting Time Zone.
When exporting data to CSV, the Export to File dialog now offers the ability to select field names that are suggested based on the query results, or to select all fields in one click.
For more information, see Exporting Data.
The Time Interval panel now displays the @ingesttimestamp/@timestamp options selected when querying events for Aggregate Alerts.
For more information, see Changing Time Interval.
A new timestamp column has been added in the Event list displaying the alert timestamp selected (@ingesttimestamp or @timestamp). This will show as the new default column along with the usual @rawstring field column.
For more information, see Alert Properties.
When a file is referenced in a query, the
Search
page now shows a new tab next to the Results and Events tabs, bearing the name of the uploaded file. Activating the file tab will fetch the contents of the file and will show them as aTable
widget. Alternatively, if the file cannot be queried, a download link will be presented instead.For more information, see Creating a File.
Sections can now be created inside dashboards, allowing for grouping relevant content together to maintain a clean and organized layout, making it easier for users to find and analyze related information. Sections can contain data visualizations as well as Parameter Panels. Additionally, they offer more flexibility when using the Time Selector, enabling users to apply a time setting across multiple widgets.
For more information, see Sections.
The
Users
page has been redesigned so that the Repository and view roles are displayed in a right hand side panel which opens when a repository or view is selected. The repository and views roles panel shows the roles that give permissions to the user for the selected repository or view, together with groups that apply to them and the corresponding query prefixes.For more information, see Manage Users.
An organization administrator can now update a user's role on a repository or view from the
Users
page.For more information, see Manage User Roles.
The design of the file editor for Lookup Files has been improved. The editor is now also more responsive and has support for tab navigation.
The Client type item in the Query details tab has been removed. Previously, Dashboard was incorrectly displayed as the value for both live dashboard and alert query types.
For more information, see Query Monitor — Query Details.
In Organization settings, layout changes have been made to the
Groups
page for viewing and updating repository and view permissions on a group.UI workflow updates have been made in the
Groups
page for managing permissions and roles.For more information, see Manage Groups.
Automation and Alerts
A maximum limit of 1 week has been added on the throttle period for Filter Alerts and Standard Alerts. Any existing alert with a higher throttle time will continue to run, but when edited, lowering the throttle time to 1 week at most will be required.
Standard Alerts have been renamed to Legacy Alerts. It is recommended using Filter Alerts or Aggregate Alerts alerts instead of legacy alerts.
For more information, see Alerts.
The
{action_invocation_id}
message template has been added: it contains a unique id for the invocation of the action that can be correlated with the activity logs.For more information, see Message Templates and Variables, Monitoring Alert Execution through the humio-activity Repository.
It is no longer possible to use @id as throttle field in filter alerts, as this has no effect. Any existing filter alerts with @id as throttle field will continue to run, but the next time the filter alert is updated, the throttle field must be changed or removed.
For more information, see Field-Based Throttling.
Audit logs for Alerts and Scheduled Searches now contain the package, if installed from a package.
The following UI changes have been introduced for alerts:
The Alerts overview page now presents a table with search and filtering options.
An alert-specific version of the
Search
page is now available for creating and refining your query before saving it as an alert.The alert's properties are opened in a side panel when creating or editing an alert.
In the side panel, the recommended alert type to choose is suggested based on the query.
For aggregate alerts, the side panel allows you to select the timestamp (@ingesttimestamp or @timestamp).
For more information, see Creating Alerts, Alert Properties.
A new Disabled actions status is added and can be visible from the
Alerts
overview table. This status will be displayed when there is an alert (or scheduled search) with only disabled actions attached.For more information, see Alerts Overview.
Audit logs for Filter Alerts now contain the language version of the alert query.
A new aggregate alert type is introduced. The aggregate alert is now the recommended alert type for any queries containing aggregate functions. Like filter alerts, aggregate alerts use ingest timestamps and run back-to-back searches, guaranteeing at least once delivery to the actions for more robust results, even in case of ingest delays of up to 24 hours.
For more information, see Aggregate Alerts.
The following adjustments have been made for Scheduled PDF Reports:
If the feature is disabled for the cluster, then the
menu item under will not show.If the feature is disabled or the render service is in an error state, users who are granted with the
ChangeScheduledReport
permission and try to access, will be presented with a banner on theScheduled reports
overview page.The permissions overview in the UI now informs that the feature must be enabled and configured correctly for the cluster, in order for the
ChangeScheduledReport
permission to have any effect.
Users can now see warnings and errors associated to alerts in the
Alerts
page opened in read-only mode.
GraphQL API
The new environmentVariableUsage() GraphQL API has been introduced for listing non-secret environment variables used by a node. This is intended as an aid to help do configuration discovery when managing a large number of LogScale clusters.
The getFileContent() and newFile() GraphQL endpoint responses will change for empty files. The return type is still
UploadedFileSnapshot!
, but the lines field will be changed to return[]
when the file is empty. Previously, the return value was a list containing an empty list[[]]
. This change applies both for empty files, and when the provided filter string doesn't match any rows in the file.The log line containing
Executed GraphQL query
in the humio repository, that is logged for every GraphQL call, now contains the name of the mutations and queries that are executed.The new concatenateQueries() GraphQL API has been introduced for programmatically concatenating multiple queries into one. This is intended to eliminate errors that might occur if queries are combined naively.
The preview tag has been removed from the following GraphQL mutations:
DeleteIngestFeed
resetQuota
testAwsS3SqsIngestFeed
The stopStreamingQueries() GraphQL mutation is no longer in preview.
The getFileContent()GraphQL query will now filter CSV file rows case insensitively and allow partial text matches. This happens when filterString input argument is provided. This makes it possible to search for rows without knowing the full column values, and while ignoring the case.
The defaultTimeZone GraphQL field on the
UserSettings
GraphQL type no longer defaults to the organisation default time zone if the user has no default time zone set. To get the default organization time zone through the API, use the defaultTimeZone field on theOrganizationConfigs
GraphQL type.The new startFromDateTime argument has been added to s3ConfigureArchiving GraphQL mutation. When set, S3Archiving does not consider segment files that have a start time that is before this point in time. This in particular allows enabling S3 archiving only from a point in time and going forward, without archiving all the older files too.
A new field named searchUsers has been added on the group() output type in graphql, which is used to search users in the group. The field also allows for pagination, ordering and sorting of the result set.
Storage
An alternative S3 client is now available and enabled by default. It handles file uploads more efficiently, by setting the Content-MD5 header during upload thus allowing S3 to perform file validation instead of having LogScale do it via post-upload validation steps. This form of validation should work for all uploads, including when server-side encryption is enabled. The new S3 client only supports this validation mode, so setting the following variables will have no effect:
In case of issues, the S3 client can be disabled by setting
USE_AWS_SDK=false
, which will set LogScale back to the previous default client. Should you need to do this, please reach out to Support to have the issue addressed, because the previous client will be deprecated and removed eventually.Support for bucket storage upload validation has changed. LogScale now supports the following three validation modes:
Checking the ETag HTTP response header on the upload response. This mode is the default, and can be opted out of via the
BUCKET_STORAGE_IGNORE_ETAG_UPLOAD
configuration parameter.Checking the ETag HTTP response header on a HEAD request done for the uploaded file. This is the second preferred mode, and can be opted out of via the
BUCKET_STORAGE_IGNORE_ETAG_AFTER_UPLOAD
configuration parameter.Downloading the file that was uploaded, in order to validate the
checksum
file. This mode is enabled if neither of the other modes are enabled.
Previous validation modes that did not compare checksums have been removed, as they were not reliable indicators of the uploaded file integrity.
The size of the queue for segments being uploaded to bucket storage has been increased. This reduces how often a scan global for changes is needed.
For more information, see Bucket Storage.
For better efficiency, more than one object is now deleted from Bucket Storage per request to S3 in order to reduce the number of requests to S3.
Support is implemented for returning a result over 1GB in size on the
queryjobs
endpoint. There is now a limit on the size of 8GB of the returned result. The limits on state sizes for queries remain unaltered, so the effect of this change is that some queries that previously failed returning their results due to reaching 1GB, even though the query completed, now work.
API
Support for array and object handling in the
fields
object has been added for Ingesting with HTTP Event Collector (HEC) events.
Configuration
A new dynamic configuration variable
GraphQlDirectivesAmountLimit
has been added to restrict how many GraphQL directives can be in a query. Valid values are integers from 5 to 1,000. The default value is 25.The
QueryBacktrackingLimit
feature is now enabled by default. The default value for the max number of backtracks (number of times a single event can be processed) a query can do has been reduced to2,000
.Adjusted launcher script handling of the
CORES
environment variable:If
CORES
is set, the launcher will now pass-XX:ActiveProcessorCount=$CORES
to the JVM. IfCORES
is not set, the launcher will pass-XX:ActiveProcessorCount
to the JVM with a value determined by the launcher. This ensures that the core count configured for LogScale is always same as the core count configured for internal JVM thread pools.-XX:ActiveProcessorCount
will be ignored if passed directly via other environment variables, such asHUMIO_OPTS
. Administrators currently configuring their clusters this way should remove-XX:ActiveProcessorCount
from their variables and setCORES
instead.The default
retention.bytes
has been modified for global topic from 1 GB to 20 GB. This is applied only when the topic is being created by LogScale initially. For existing clusters you should raise retention on the global topic so that it has room for at least a few hours of flow. This is only relevant for large clusters, as small clusters do not produce enough to exceed 1 GB per few hours. It is ideal to have room for at least 1 day in the global topic for better resilience against large spikes in traffic combined with losing global snapshot files.Cluster-wide configuration of S3 Archiving is introduced, in addition to the existing repo-specific configurations. This feature allows the cluster admin to setup archiving to a (single) bucket for a subset of repositories on the cluster, fully independent of the S3 Archiving available to end users via the UI. This feature adds the following new configuration parameters:
S3_CLUSTERWIDE_ARCHIVING_ACCESSKEY
(required)S3_CLUSTERWIDE_ARCHIVING_SECRETKEY
(required)S3_CLUSTERWIDE_ARCHIVING_REGION
(required)S3_CLUSTERWIDE_ARCHIVING_BUCKET
(required)S3_CLUSTERWIDE_ARCHIVING_PREFIX
(defaults to empty string)S3_CLUSTERWIDE_ARCHIVING_PATH_STYLE_ACCESS
(default isfalse
)S3_CLUSTERWIDE_ARCHIVING_KMS_KEY_ARN
S3_CLUSTERWIDE_ARCHIVING_ENDPOINT_BASE
S3_CLUSTERWIDE_ARCHIVING_WORKERCOUNT
(default iscores/4
)S3_CLUSTERWIDE_ARCHIVING_USE_HTTP_PROXY
(default isfalse
)S3_CLUSTERWIDE_ARCHIVING_IBM_COMPAT
(default isfalse
)
Most of these configuration variables work like they do for S3 Archiving, except that the region/bucket is selected here via configuration, and not dynamically by the end users, and also that the authentication is via explicit accesskey and secret, and not via IAM roles or any other means.
The following dynamic configurations are added for this feature:
S3ArchivingClusterWideDisabled
(defaults tofalse
when not set) — allows temporarily pausing the archiving in case of issues triggered by, for example, the traffic this creates.S3ArchivingClusterWideEndAt
andS3ArchivingClusterWideStartFrom
— timestamps in milliseconds of the "cut" that selects segment files and events in them to include. When these configuration variables are unset (which is the default) the effect is to not filter by time.S3ArchivingClusterWideRegexForRepoName
(defaults tonot match
if not set) — the repository name regex must be set in order to enable the feature. When set, all repositories that have a name that matches the regex (unanchored) will be archived using the cluster wide configuration from this variable.
Ingestion
On the Code page accessible from the menu when writing a new parser, the following validation rules have been added globally:
Arrays must be contiguous and must have a field with index 0. For instance,
myArray[0] := "some value"
Fields that are prefixed with
#
must be configured to be tagged (to avoid falsely tagged fields).
An error is displayed on the parser Code page if the rules above are violated. This error will not appear during actual parsing.
For more information, see Creating a New Parser.
To avoid exporting redundant fields in the parsers, LogScale will now omit YAML fields with a
null
value when exporting YAML templates — even when such fields are contained inside a list. Omitting fields with anull
value previously only happened for fields outside a list.
Log Collector
RemoteUpdate
version dialog has been improved, with the ability to cancel pending and scheduled updates.
Functions
Matching on multiple rows with the
match()
query function is now supported. This functionality allowsmatch()
to emit multiple events, one for each matching row. Thenrows
parameter is used to specify the maximum number of rows to match on.For more information, see
match()
.The
match()
function now supports matching on multiple pairs of fields and columns.For more information, see
match()
.The new query function
text:contains()
is introduced. The function tests if a specific substring is present within a given string. It takes two arguments:string
andsubstring
, both of which can be provided as plain text, field values, or results of an expression.For more information, see
text:contains()
.The new query function
array:append()
is introduced, used to append one or more values to an existing array, or to create a new array.For more information, see
array:append()
.
Fixed in this release
Falcon Data Replicator
Testing new FDR feeds using s3 aliasing would fail for valid credentials. This issue has now been fixed.
UI Changes
The
Query Monitor
page would show queries running on @ingesttimestamp as running on a search interval over all time. This wrong behavior has been fixed to show the correct search interval.The event histogram would not adhere to the timezone selected for the query.
When managing sessions within an organization, it was not possible to sort active sessions by the Last active timestamp column. This issue has now been fixed.
In the Export to File dialog, when using the keyboard to switch between options, a different item than the one selected was highlighted. This issue has now been fixed.
A long list of large queries would break the queries' list appearing under the Recent tab by not being updatable. The limit to recent queries has now been set to 30.
For more information, see Recalling Queries.
A race condition in LogScale Multi-Cluster Search has been fixed: a
done
query with an incomplete result could be overwritten, causing the query to never complete.The dialog to quickly switch to another repository would open when pressing the undo hotkey on Windows machines. This wrong behavior has now been fixed.
The dropdown menu for selecting fields used when exporting data to a CSV file was hidden behind the Export to file dialog. This issue has now been fixed.
The
Organizations
overview page has been fixed as the Volume column width within a specific organization could not be adjusted.The display of Lookup Files metadata in the file editor for very long user names has now been fixed.
The settings used to disable automatic searching would not be respected when creating a new alert. This issue has now been fixed.
When Creating a File, saving an invalid
.csv
file was possible in the file editor. This wrong behavior has now been fixed.The Export to file dialog used when Exporting Data has been fixed as CSV fields input would in some cases not be populated with all fields.
Fixing a visualization issue where the values in a multi-select combo box could overlap with the number of selected items.
When clicking to sort the Sessions based on Last active, the sorting was wrongly based on Login time instead. This issue has now been fixed.
It was not possible to sort by columns other than ID in the Cluster nodes table under the UI menu. This issue has now been fixed.
Automation and Alerts
Actions would show up as scheduled searches and vice versa when viewing the contents of a package. This issue has now been fixed.
Fixed an issue where queries that were failing would never complete. This could cause Alerts and Scheduled Searches to hang.
Scheduled Searches would not always log if runs were skipped due to being behind. This issue has been fixed now.
The read-only alert page would wrongly report that actions were being throttled when a filter alert had disabled throttling. This issue has now been fixed.
GraphQL API
The getFileContent() GraphQL endpoint will now return an
UploadedFileSnapshot!
datatype with the field totalLinesCount: 0 when a file has no matches for a given filter string. Previously it would return the total number of lines in the file.The background processing underlying the redactEvents() mutation would fail if the filter included tags. This error has now been fixed.
Storage
Throttling for bucket uploads/downloads has been fixed as it could cause unintentionally high number of concurrent uploads or downloads, to the point of exceeding the pool of connections.
Notifying to Global Database about file changes could be slow. This issue has now been fixed.
Segments could be considered under-replicated for a long time leading to events being retained in Kafka for extended periods. This wrong behavior has now been fixed.
Throttling for bucket uploads/downloads could cause unintentionally harsh throttling of downloads in favor of running more uploads concurrently. This issue has now been fixed.
Digest threads could fail to start digesting if
global
is very large, and if writing toglobal
is slow. This issue has now been fixed.The throttling for segment rebalancing has been reworked, which should help rebalancing keep up without overwhelming the cluster.
API
fields
object did not show up in Ingesting with HTTP Event Collector (HEC) events. This issue has now been fixed.
Configuration
Make a value of
1
forBucketStorageUploadInfrequentThresholdDays
dynamic configuration result in all uploads to bucket being subject to "S3 Intelligent-Tiering". Some installs want this as they apply versioning to their bucket, so even though the life span as a non-deleted object is short, the actual data remains for much longer in the bucket, and then tiering all objects saves on cost of storage for them. Objects below 128KB are never tiered in any case.
Ingestion
Event Forwarding would fail silently if an error occurred while executing the query. This issue has now been fixed.
A queryToRead field has been added to the filesUsed property of queryResult to read the data from a file used in a query.
For more information, see Polling a Query Job.
Event Forwarding using
match()
orlookup()
with a missing file would continue to fail after the file was uploaded. This issue has now been fixed.When shutting down a node, the process that load files used by a parser would be stopped before the parser itself. This could lead to ingested events not being parsed. This issue has now been fixed.
A wrong order of the output events for parsers have been fixed — the output now returns the correct event order.
Dashboards and Widgets
Arguments for parameters no longer used in a deleted query could be submitted anyway when invoking a saved query that uses the same arguments, thus generating an error. This issue has now been fixed.
The
Table
widget has been fixed due to its header appearing transparent.
Log Collector
Queries that were nested too deeply would crash LogScale nodes. This issue has now been fixed.
Functions
parseXml()
would sometimes only partially extract text elements when the text contained newline characters. This issue has now been fixed.Parsing the empty string as a number could lead to errors causing the query to fail (in
formatTime()
function, for example). This issue has now been fixed.The query backtracking limit would wrongly apply to the total number of events, rather than how many times individual events are passed through the query pipeline. This issue has now been fixed.
Long running queries using
window()
could end up never completing. This issue has now been fixed.writeJson()
would write invalid JSON by not correctly quoting numbers starting with unary plus or ending with a trailing.
(dot).A regression issue has been fixed in the
match()
function incidr
mode, which was significantly slower when doing submission of the query.
Known Issues
Queries
Improvement
UI Changes
The performance of the query editor has been improved, especially when working with large query results.
Automation and Alerts
The log field
previouslyPlannedForExecutionAt
has been renamed toearliestSkippedPlannedExecution
when skipping scheduled search executions.The field
useProxyOption
has been added to Webhooks action templates to be consistent with the other action templates.The severity of a number of alert and scheduled search logs has been changed to better reflect the severity for users.
Storage
The global topic throughput has been improved for particular updates to segments in datasources with many segments.
For more information, see Global Database.
Let segment merge span vary by +/- 10% of the configured value to avoid all segment targets switching to a new merge targets at the same point in time.
Ingestion
The input validation on Split by AWS records preprocessing when Set up a New Ingest Feed has been simplified: it will still validate that the incoming file is a single JSON object (and not, for example, multiple newline-delimited JSON objects), but the object may or may not contain a
Records
array. This resolves an ingest feed issue for CloudTrail with log file integrity enabled. In such cases, the emitted digest files (that does not have theRecords
array) would halt the ingest feed. These digest files are now ignored.For more background information, see this related release note.
The Split by AWS records preprocessing when Set up a New Ingest Feed now requires the
Records
array. This better protects against a situation where mistakenly using this preprocessing step with non-AWS records would interpret the files as empty batches of events, leading notifications in SQS to be deleted without ingesting any events.
Queries
Cache files, used by query functions such as
match()
andreadFile()
, are now written to disk for up to 24 hours after use. This can improve the time it takes for a query to start significantly, however, it naturally takes op disk space.A fraction of the disk used can be controlled using the configuration variables
TABLE_CACHE_MAX_STORAGE_FRACTION
andTABLE_CACHE_MAX_STORAGE_FRACTION_FOR_INGEST_AND_HTTP_ONLY
.