Functional Testing - LoadGen Studio functional projects now run headless on the Platform under the Functional & Automated Testing licence. A new Functional Testing section has a Dashboard, Packages, Run History, Schedules and Engines. Upload a LoadGen Studio repository on Packages to browse its projects, testcases and variables and run them; a testcase's Script button shows its steps and its generated code, and Reuse this testcase saves it as a workload for a load profile or an End-to-End monitoring profile. Schedules run testcases on a cron expression in a time zone, with a threshold, a timeout and alert profiles, and the Upcoming view lists the runs of the next seven days with Run now. Engines enrol with a token whose secret and enrol command are shown once. The Functional Testing Agent has its own tile on the Updates tab of Infrastructure > Appliance and on Settings > Software Updates, and is updated on its own.
Functional Testing dashboard and run page - the Dashboard shows result tiles (Runs passed, Failed runs, Unfinished runs, Flaky testcases, Median run time), the engines online and the run slots in use, the most frequent failures, and the unstable testcases with their last ten results and New failure, Always failing and Flaky badges. A run that failed or did not finish says why, rule by rule, with Open testcase for each failed or unfinished testcase. The run page has a Steps tab with a run timeline, a Screenshots filmstrip whose Play replays the run as a slideshow, a filtered Log, and a Download report menu (PDF report, JUnit XML, HTML summary, CSV rows) beside Rerun. Run History filters by verdict, period and text.
API flows in load tests and End-to-End Monitoring - an API Testing flow can now run as a load test or as an End-to-End Monitoring check on Core Agents. The load-profile wizard gains an API flow technology: pick the flow, a pinned or the latest committed revision and the environment, then set iterations, maximum time, pacing, think time, request timeout, data rows and what happens when an iteration fails. The monitoring wizard gains an API Flow target (Core Agents, Direct mode), where the flow is the workload and a test user is optional. In the flow editor, Use in offers Load test… (with a request-rate estimate that also counts linked flows), E2E monitoring… and Check agent readiness, and the flow's Used by list shows the profiles that use it; a flow a profile uses cannot be deleted. Core Agents gains an API flow runner column with readiness and capacity, and Accept new key when an agent's runner key changed. A failed run shows the reason and the remedy on Active Test, on Results and in the monitoring session drawer.
Bulk assign for large tests - the Users & Agents step of every load-profile wizard gets a Bulk assign dialog. Pick users by count, by a pattern such as "Prefix[Start:End:Digits]" or by pasting a list; the preview says how many were found, are missing or are already on the profile, and missing Web, Mobile & API users are created in one go. Split the users over a workload mix, then let agents be chosen automatically or pin all users to one agent, optionally activating online agents until they cover the users; a Core StoreFront profile also gets a resource picker. The step gains a capacity strip ("covers" or "N short"), a Groups view, agent reachability filters, an Activate menu ("Enough for N users", "All online agents", "Selected", "Deactivate all") and Set max slots…. The "Web users" pool is now called "Web, Mobile & API users".
Agent screenshots - the screenshots Core and Full Agents take on failed transactions and SendScreenShot actions are now kept and shown. Testing > Active Test gains a screenshots tile, a live strip of the newest screenshots and a Shots column in the user list, whose rows open a user timeline with the screenshots inline. Test Analysis gains a Screenshots tab with a screenshots-over-time chart, grouping by look-alike screen, transaction, user or agent, a gallery, Export screenshots (.zip) and, for TestAdmin, Delete screenshots. The viewer steps through screenshots with the arrow keys and shows the transaction, message, user and time. The End-to-End Monitoring session drawer gets a Screenshots section, and the new Screenshot Storage configuration page shows disk use and sets capture, retention, per-run caps, a minimum free disk space and whether backups include screenshots.
Pinned baselines and the Impact tab - Pin this run as baseline in Test Analysis, or Pin as baseline on Testing > Test Runs, sets the load profile's baseline, and every later run is scored against it. Test Analysis gains an Impact tab that compares the run with the baseline per application and measurement (baseline and run p95, failures, Apdex and a verdict), with Regressed, Improved, Unchanged, New and Missing tiles as filters and the hosts under CPU pressure; Change baseline compares with another run for this view only. Test Runs gains a Baseline column, Test Analysis > Trends opens with "LoadGen Score over time" with the pinned baseline ringed, and Testing > Load Profiles shows each profile's baseline history. The LoadGen Score panels name the baseline a score was measured against, and the report dialog suggests the pinned baseline under Compare against. A CSV or JSON datasource has no impact view.
Active Test shows the whole run and why users failed - leaving Testing > Active Test and coming back no longer starts the charts over: they are filled from the series the API keeps for the run and continue live. The page heads with the run's own name, with the load profile names as a subtitle. The User Fleet shows a failed user's reason in red with its source, and a "Why users failed" panel on the Overview tab of Test Analysis and in the run summary on Test Runs lists the reasons with counts. The Phase Timeline shows each phase as "Waiting", "Running", "Completed", "Completed with errors" or "Failed", marks the current phase "Now", and says what ends the run, for example "Run ends when phase 3's wait events have fired (0/5), or at 1h, whichever comes first."
Ingestion Pipeline Health shows what happened to the results - the panel on Testing > Active Test shows packets received, dropped, spooled to disk and replayed, and rows committed, with one banner that says whether anything was lost, from "All results of the run are committed" to "Results incomplete". While the run commits its last rows the panel says how many are pending, and the run shows Completed once they are in; SPOOLING, COMMITTING and INCOMPLETE badges join DROPS. When InfluxDB refuses large writes (HTTP 413), usually because of a proxy body limit, an amber line says LoadGen now writes smaller requests and that raising that limit (nginx: client_max_body_size) restores full speed. Rows the datasource refused one by one are set aside for an administrator with the reason per row, and the panel says so.
Core load profiles for command lines and Citrix StoreFront - a Core load profile runs a command line or a Citrix StoreFront session on the Core Agent. For StoreFront the wizard saves the StoreFront address and resource, and each test user signs in with their own user name and password. For a command line the program, arguments and working directory are saved together, and {username}, {domain} and {password} in the arguments are replaced with each test user's values. Core profiles ask for no workload ("Workload: Not used"). A StoreFront profile imported from LoadGen Director opens on StoreFront with its address and resource.
Web Testing and Command Line run on Core Agents only - the monitoring wizard no longer offers Full Agents for Web Testing and Command Line, and the load-profile wizard none for Web, Core, Mobile and API flow load tests. A stored profile that still holds a Full Agent shows it in red with a one-click swap to the Core Agent on the same host ("Use the Core Agent "), and the agent legend says what each agent kind runs. The wizards also flag an agent that was deleted, and when an agent was deleted and added again under the same name they offer to switch to it, or switch by themselves and say so. The Review step lists every agent a save leaves out, and why.
Load-profile wizards offer only settings that take effect - the RDP wizard no longer offers port, Network Level Authentication, encryption, compression, full screen or session settings, and says "The RDS worker always connects on port 3389 with Network Level Authentication and compression on; they cannot be changed per profile." The WVD wizard's session and display settings are replaced by a note that the session uses the host pool's RDP properties. The Citrix wizard shows audio, compression, encryption, session reliability and persistent cache only on a Direct / Manual connection, and its display overrides are labelled "Override the ICA file's display settings with profile values" and now take effect, StoreFront included. A setting switched off, cleared or set back to its default now stays that way after saving, and values set in the Advanced settings table are kept.
Citrix ICA file upload - in ICA file mode the Citrix load-profile and monitoring-profile wizards take the .ica file itself instead of an "ICA File Path", which no agent read. The file is stored with the profile and shown with its size and the application it opens, with Replace and Remove, and its Password and ClearPassword lines are removed on save. An empty file, a file over 65,536 characters or a file without an [ApplicationServers] entry is refused with the reason.
Citrix External Login - choosing External Login in the Citrix load-profile wizard now switches the mode on, and a saved External Login profile reopens as External Login. Fetch signs in through the appliance browser with a test user or manual credentials and fills the Application Name list; when it cannot, it says why in one sentence. The script toolbar offers Download user scripts, Download shipped scripts and Upload user scripts, and user scripts win by name. On an installation without the appliance browser, the wizard shows an amber notice that the sign-in check runs only on the LoadGen Appliance, and the test still runs in the agent's browser.
Insight > Experience - a new page shows the experience of real users per machine group, for the latest scored window: tiles for the UX score against its 7-day baseline, the application launch p95 against its baseline, the input delay p95 and the hosts under pressure; a banner for the worst regression; an Applications table, worst first, with score, launch p95, sessions, a 24-hour sparkline and failures, timeouts and crashes; and a Hosts table. Recompute baseline now is offered to roles allowed to recompute it.
Insight alerts, Service Level alerts and Syslog - the alert rule editor gains the Insight scope, with Windows in a row, notify "Once per regression, then when it recovers" or "Every matching window", and a quick setup on "Score drops more than (points)", "Launch p95 above baseline by more than (%)" or "Crashes more than". Service Levels alert profiles can now be created. Triggers gain a Syslog type with collector host, port, UDP or TCP, framing, facility, app name and a message template. A new Configuration > Alerting page holds the allow-list of internal webhook and syslog targets (exact hosts, IP addresses and CIDR ranges).
AI features switch - a new Settings > Platform tab holds the "AI features" card: Enabled, Disabled, or "Disabled · locked by administrator" when the LOADGEN_AI_DISABLED environment variable decides, with who set it and when. Only a global admin gets the toggle, and a change applies at once. While AI is disabled no AI control is shown and no AI call is made: Ask AI, the Analyse Copilot, the AI draft in report templates, Generate with AI and AI Suggest Nodes in API Testing, Settings > AI Providers, "AI Credits & Billing" and the "Weekly AI Insights" toggle in SessionSight are all hidden.
MCP client key - every MCP client now sends the MCP client key with each request. A global admin manages it on the MCP Server card of the Services tab of Infrastructure > Appliance: Show key, Copy, Rotate (which asks first) and, when there is no key yet, Create key. Every reveal and rotation is audited. The MCP Server card cannot be enabled while AI features are disabled, and says why.
LoadGen Cloud in the menu - a new LoadGen Cloud sidebar group holds Account, LoadGen Exchange, Users, Licensing (moved here from Infrastructure; old links still open it) and AI Credits & Billing. On LoadGen Cloud > Account you sign in with your own LoadGen Cloud e-mail address and password, see your organisation, what you may do and when the sign-in expires, and reset your password. LoadGen Cloud > Users (GlobalAdmin) lists the organisation's users with one licence pill per product, and creates, edits, resets, activates, deactivates and deletes users. A Licences tab lists every licence of the organisation with product, person, machine and end date. The header's cloud icon shows the LoadGen Cloud sign-in, and the proxy self-test names exchange.loadgen.cloud, platform.loadgen.cloud and licensing.loadgen.cloud for your proxy allow-list.
LoadGen Exchange on the appliance - the Workload Exchange is now called the LoadGen Exchange throughout the interface. LoadGen Cloud > LoadGen Exchange browses the Exchange by kind, with search and Audience, Technology and Tag filters; each card shows the publisher, the technology, the audience, the version and whether the item runs code, with Import or Update to vN. The import dialog warns when an item runs code on your appliance and asks for parameters, a credential profile per credential slot and a conflict policy. Publish to the LoadGen Exchange… is on Uptime > Checks, API Testing > Flows, the Analyse dashboards, saved charts and report templates, counter templates, hypervisor templates and Configuration > Workloads, whose Exchange column shows Published, Update available, From the Exchange or Not published. My published shows each item's review state.
API Exposure uptime check - a new check type on Uptime > Checks calls each listed API endpoint without credentials (GET or HEAD only, no cookies, no redirects) and fails when one answers with data instead of 401 or 403. The wizard builds the list of up to 1,000 endpoints from an OpenAPI document, Scan web app or Import HAR, next to Expected public paths and a Never call list, and can refresh the list before every run. The Exposure report row action shows Protected, Exposed, Inconclusive, Not found and Public tallies, the endpoints new or gone since the previous run, a trend, and the evidence per endpoint. The Uptime overview shows the exposed and newly exposed counts, and alert triggers can use them. The Uptime > Checks filter now has a Security group (API Exposure) and a DevOps group (Repository Health).
Hyper-V hosts - a SUT profile gets Add Hyper-V host, which opens the machine drawer as a Hyper-V host with the built-in "Hyper-V host" counter template. The drawer has Host memory (GB) and Check counters, which lists the counters the score and charts read as found or missing and explains "access denied" and "network path not found". An ordinary machine that collects Hyper-V counters gets a banner with Convert to Hyper-V host.
KVM, OpenShift Virtualization and a simulated vCenter - Configuration > Hypervisor Connections offers "KVM (Prometheus)" and "OpenShift Virtualization", which take a Prometheus URL and no credential, a Basic or a Bearer credential profile; a credential on a plain http:// address is flagged before saving. The new "Simulated (demo)" type is a simulated vCenter that contacts nothing and gives three hosts, one per load level (healthy, busy, overloaded); the Hypervisor score says its data is simulated.
Counter templates - Configuration: Systems Under Test > Templates marks built-in templates with a "Built-in" badge and version; they update with the Platform and open read-only, and Copy makes an editable copy. The editor flags counters not yet confirmed on a real host, says how a wildcard instance combines, shows the localised counter name, and has a Named instance field on a copy of the SQL Server template. The new Hypervisor Templates page creates and edits hypervisor counter templates and applies them to hosts with Apply to hosts….
Counter Units as a worklist - Counter Units opens with clickable tiles (Counters, Suggested, Set by you, Not declared) that filter the table, groups the rows per counter category, adds a search box, and keeps Save and Discard in the toolbar beside an "N unsaved changes" chip.
Launch probe catalogue - on Insight > Machine Groups, the Launch Probes of Foreground Window Monitoring offer a catalogue of common applications that adds a probe in one click: Microsoft 365 (Word, Excel, PowerPoint, Outlook classic, OneNote, Visio, Access), Edge, Chrome, Firefox, Acrobat / Reader and Notepad, with Add all per group. Hints under a row say when a save will be refused or when an earlier probe already matches the process.
Trial test limit - on a trial licence, Testing > Run Test shows "Trial licence: tests stop after 15 minutes." above Start, Testing > Active Test shows the time left, and a run ended by the limit shows a blue "Trial limit:" notice on Results instead of an error. A test planned longer than the limit is refused at start with the planned and the allowed minutes.
Replace only the User Data Secret - the Security Tokens tab of Settings and the Security tab of Infrastructure > Appliance ask what to replace: All tokens, as before, or Only the User Data Secret. The help says that existing encrypted data stays readable and that User Data Key Rotation re-encrypts it.
[#9758] A static appliance can now be switched back to DHCP from the Network tab of Infrastructure > Appliance, whose IP dialog offers Static address or Automatic (DHCP) instead of failing with "Interface, Address, and Gateway are required.", and from the Services Setup Wizard, whose "Use DHCP (automatic)" now switches the network as the last step of Apply. A "Switching to DHCP" banner counts down while the appliance waits for a lease: it keeps DHCP once it holds one, and returns to its previous settings by itself when no DHCP server answers within 60 seconds.
[#9759] Cards, tables and long values now fit a 1024-pixel tablet or a 14-inch laptop across the app. On the Updates tab of Infrastructure > Appliance the component tiles show their full names without blank blocks and the archive path wraps with a copy button, and the row actions on the Alert Profiles and Alert Triggers pages stay within reach. Long paths, URLs, host names and error lines wrap instead of being cut, wide tables scroll inside their card with the Actions column pinned, and forms keep a readable width on very wide screens.
Fixed the Users & Agents step of the load-profile wizards. Available Users and Assigned to Profile scroll inside a fixed height, the header check box selects the visible page and offers "Select all N matching users", and Add, Remove and the batch edits act only on selected users the filter shows, so users hidden by a filter are no longer removed. Each user's own resource is saved with the profile.
Fixed Testing > Active Test freezing for the rest of the run after one slow answer, and showing red on a healthy run: the agent utilization bar is blue at every fill level, and charts draw the Failed line only once something has failed.
Fixed Testing > Run Test keeping the "already running" banner and a disabled Start Test after the run had finished.
Fixed Test Analysis of a run that is still running showing "Test not found."; it now says the run is still running and offers Open Active Test.
Fixed the Updates tab of a Launcher Appliance offering no push button for a build picked from the archive when the component already ran a newer build. A picked build now shows Push whenever it is not the build already running.
Fixed the Services Setup Wizard giving up while InfluxDB and Grafana were still starting. It now waits, and its InfluxDB, default datasource and Grafana steps retry while a service is still starting.
Fixed delete buttons, such as the one on Alerting > Profiles, deleting on the first click. Alert profiles, alert triggers, SSO providers, status page incidents and maintenance windows, the GA4 connection, replay notes, appliance readiness profiles and SUT machines now ask "Delete ''? This cannot be undone." first.
Fixed deleting an agent that profiles use saying only "Cannot delete!". The dialog now names the profiles, for example "'X' is used by 2 monitoring profiles (A, B) and 1 load profile (C). Reassign those profiles to another agent first, then delete it."
Fixed saving an alert profile resetting a notify-once profile to notify on every match with recovery off.
Fixed Command Line monitoring profiles not saving their command. The wizard saves the command path, which it requires, and new Arguments and Working Directory fields take {username}, {domain} and {password}. A Command Line profile needs no workload.
Fixed a load-profile save refused by the API showing raw JSON: each rule now shows as a sentence on the step it belongs to. The Review step names the SUT profile, the API flow, the environment and the datasource instead of ids, and choosing "-- None --" clears a SUT profile link.
Fixed Uptime counters counting disabled checks as Healthy. The Uptime and Monitoring overviews count enabled checks only and show paused checks in a grey pill, a check that never ran reads "Never", and a result written by another installation is labelled "Other installation".
Fixed Add Agent and Edit Agent hiding the API's validation errors: an invalid address is shown under the address field, and other refusals in the banner.
Fixed drop-downs showing "-- choose --" or their first entry instead of the stored value when their list loaded late, for example the flow on Use in > Load test, the environment and auth profile in the API Testing run dialog, Windows Run-As when editing an agent, and the LoadGen Exchange import version.
Fixed the test run picker showing a datasource whose run list could not be read as empty: an amber line now gives the reason.
Fixed a bulk import on Configuration: Testing > Users ending without feedback; a banner now says how many users were imported and how many already existed.
Fixed the API Testing flow editor header squeezing the flow name on narrower windows, the flow wizard saying "No versions available" while it was still reading, and Import flow from JSON reopening with the previous package, which could create a duplicate flow.
When LoadGen Cloud refuses the appliance's licence token, the LoadGen Exchange and LoadGen Cloud pages say "LoadGen Cloud refused this appliance's licence token" and to contact LoadGen support, instead of blaming the stored account.
Fixed dates and numbers following the Windows host's language: dates read in English ("18 Oct 10:22") and numbers as "1,234.5" on every host. Ramp chart axes no longer repeat their labels, durations read "45 s" or "1 min 30 s", and counts read "1 user" and "2 users".
Fixed buttons drawing at the wrong size or as bare text, icons that never appeared (for example on Run, Resume, Pause, Upload and Reboot buttons), KPI tiles drawing a bare black icon, garbled characters in a few hints and lists, and the page scrolling behind a wizard. Screens say "Agent" instead of "LoadBot".
The System Status dropdown reads the host's details with the signed-in session; a figure that is missing shows "-".
Requires the API from release 1.0.0.12864 or later.