Release Notes
Release Notes — 2.77
Covers the 2.77 release series. Most recent patch: 2.77.4.
Unless a subsection notes otherwise, changes apply to both Monad SaaS and self-hosted.
New features & improvements
SaaS & self-hosted
- Pipeline edits restart only the nodes you changed — saving a pipeline used to restart every node in it, pausing ingestion across the whole pipeline. Each node now carries its own configuration fingerprint, so a change to one node's settings or edges rolls that node alone; rotating a secret still rolls the nodes that use it. The first save after the upgrade restarts each pipeline once as nodes pick up their own fingerprints. (2.77.1)
- Monad MCP — inspect a node's retry queue — the new
list_retry_queuetool returns the records a node has failed to deliver together with the error the destination returned, which is the fastest answer to "the pipeline is running but nothing arrives at my output." Payloads are omitted unless you ask for them, and reading the queue never disturbs live processing. Docs → (2.77.1) (For teams using the Monad MCP integration.) - Monad MCP — the community transforms catalog is live —
list_community_transformsnow reads Monad's live community-transformations repository (hundreds of CEF, ECS and OCSF transforms, refreshed hourly) instead of a small fixed list, and can be searched by the input type a transform targets. The newget_community_transformreturns one transform's configuration, ready to pass tocreate_transform. (2.77.1) (For teams using the Monad MCP integration.) - Deduplicate and Hash transform operations — two new operations for transforms. Hash writes a digest of the whole record or of selected fields to a key you choose (SHA-256 by default; MD5, SHA-1 and xxh128 are also available), giving you a stable fingerprint for correlation. Deduplicate drops repeat records seen within a rolling window of up to five minutes, keyed on the whole record or on selected fields. Deduplicate docs → · Hash docs → (2.77.1)
- Access: opt-in — add a Hash or Deduplicate operation to a transform. Deduplicate compares across records, so it must be the only operation in the transform or the last one.
- AWS CloudWatch Logs: Lookback Window for late-arriving events — a new Lookback Window (seconds) setting on the AWS CloudWatch Logs and Amazon EKS Audit Logs inputs re-reads a trailing window on every poll.
Events that reach CloudWatch after the input has moved past their timestamp are still collected.
The docs include a CloudWatch Logs Insights query for sizing the window. Docs → (2.77.3)
- Access: opt-in — set Lookback Window (seconds) in the input's settings (5 seconds to 14 days). Leaving it unset keeps the current behavior.
- Splunk HEC input accepts Basic authentication — senders that authenticate with Basic auth, such as Tailscale log streaming, can now deliver to the Splunk HEC input.
The token is read from the password, and a leading
Splunkis ignored. Docs → (2.77.3) - Component edit pages name the component you are editing — the edit page now shows a subtitle with the component instance's name. (2.77.3)
Self-hosted only
- Cluster Sizing guide with an interactive calculator — a new guide for self-managed deployments estimates the cluster you need from your pipeline workload: enter the shape of your pipelines and your node size, and it returns node count, total CPU, memory and pods, the binding constraint, and the matching values.yaml guidance. Docs → (2.77.1)
- Helm install notes point at your console — after
helm installorhelm upgrade, the chart now prints the address of your Monad console, the sign-in steps for the authentication mode you run (bootstrap administrator credentials for local login, or your identity provider), and the other published endpoints, instead of a generic placeholder. (2.75.1) - NATS authentication across the platform — the new
global.monad.natsCredsSecretvalue names one Secret that holds your NATS credentials. Every component that talks to NATS mounts it, and the pipeline operator passes it on to the pods it creates for pipelines, scheduled inputs and alerting. Docs → (2.77.2)- Access: opt-in — create a Secret with a
nats.credskey and setglobal.monad.natsCredsSecretin your values override. Leaving it empty (the default) changes nothing. A pod does not start while the named Secret or its key is missing.
- Access: opt-in — create a Secret with a
New connectors
- OpenAI Enterprise Compliance Logs (input) — collects ChatGPT Enterprise Compliance Logs Platform data for a ChatGPT workspace or an API Platform organization: audit actions, user authentication, conversations and messages, Codex usage, plugin activity and more, delivered by OpenAI as time-windowed JSONL files and emitted by Monad as individual records for security monitoring, compliance reporting and e-discovery. Choose the source and the event-type categories your compliance API key is scoped for. Docs → (2.77.1)
- Tailscale Network Flow Logs (input) — Beta — collects the per-node traffic telemetry Tailscale records for Premium and Enterprise tailnets: virtual, subnet, exit-node and physical traffic counts between nodes. You can backfill up to Tailscale's 30-day retention. As an alternative to polling, Tailscale can push the same logs to a Splunk HEC input through log streaming. Docs → (2.77.3)
- Tailscale Audit Logs (input) — Beta — now listed in the connector catalog. It collects configuration changes, user actions and administrative events across your tailnet. Docs → (2.77.3)
- Finxact Login (input) — Beta — takes a full snapshot of every login on a Finxact tenant on each run: the associated user and party, knowledge-based-authentication questions, biometrics and other credential posture, so changes to existing logins are always captured. Docs → (2.77.3)
- Finxact Login Events (input) — Beta — collects Finxact's login event logs incrementally. Choose Login Activity for authentication events or Login History for credential changes, and add a second input to collect both. Docs → (2.77.3)
Fixes
- Monad MCP —
create_inputandcreate_outputno longer drop the connector configuration — passing the configuration asconfigwas silently stripped, so any connector with a required setting could not be created;configis now accepted alongsideconfig_json, and every tool refuses an unknown argument with a hint at the nearest valid one instead of ignoring it. (2.77.1) - Alerting keeps working after an organization's last pipeline is removed — deleting every pipeline in an organization and later adding a new one left that organization without alert evaluation until an alert rule was next edited; the alerting service is now recreated automatically. (2.77.1)
- AWS SQS S3 inputs ignore S3 test events — the
s3:TestEventnotification that S3 sends when a bucket notification is first configured no longer surfaces as a pipeline error on a new SQS S3, CloudTrail or GuardDuty input; it is skipped. (2.77.1) - Google Cloud Storage input accepts its default Partition Format — the default value shown for Partition Format was rejected by the input's own validation, so a new input could not be saved without changing it; the default now validates. (2.77.1)
- Box User Activity input reads Box's streaming events feed — the input now uses the
admin_logs_streamingfeed Box recommends instead of the delayedadmin_logsfeed, so events arrive without Box's unspecified lag; existing inputs migrate automatically. (2.77.1) - Clumio Consolidated Alerts input delivers alert updates — the input keyed on when an alert was first raised, so later updates to an already-collected alert were never fetched; it now tracks the alert's last update and picks up changes. (2.77.1)
- Braintrust Audit Logs input no longer misses late-published rows — the query now holds back a short window so a row that becomes queryable after its creation time is still collected on a later run. (2.77.1)
- Node details panel no longer jumps on high-throughput pipelines — record counts of a million or more are shown compactly (for example
2.4M rec) so the panel keeps a steady height as counters tick. (2.77.1) - Select all / Deselect all sit on the field's label line in multiselect settings fields instead of floating above it. (2.77.1)
- Sign in with SSO is readable in light mode — the workspace-picker buttons on the sign-in card rendered black text on a dark background in light mode. (2.77.1)
- Palo Alto Cloud Security Alerts input no longer skips events after a pause — the relative lookback window is now rounded up, so events from the first minute after a pipeline is re-enabled are collected instead of falling into a gap. (2.77.1)
- Finxact Messages and Error Log inputs no longer stall on a missing detail record — when Finxact returns no detail for a message or error entry, the input emits the entry on its own and moves on instead of retrying the same record indefinitely. (2.77.1)
- Google Workspace Alert Center input authenticates with the Alert Center scope — the input requested the Admin Reports scope instead of
apps.alerts, so authentication failed even when the delegation matched the docs. (2.77.3) - Google Workspace activity inputs' Test Connection checks your credentials — on a new input, the test reported success without calling Google, so bad credentials passed. It now makes a real authenticated request. (2.77.3)
- CISA KEV and GitHub Advisory Database inputs pass Test Connection again. (2.77.3)
- Output nodes report Running after recovering from a startup error — an output whose first setup attempt failed stayed Erroring until data flowed, even after it recovered. (2.77.3)
- Alerts list no longer comes back empty while alerts exist — listing alerts could return an empty result during a brief startup race. It now returns the alerts, or a retryable error. Billing alerts in the list now require billing read permission, as they already did in the live alert stream. (2.77.3)
- API key expiry notifications are no longer lost — a transient delivery failure dropped the 30-, 7- and 1-day expiry alerts for good. They are now retried until delivered. (2.77.3)
- Clearer API errors — creating a secret with a name that already exists returns
409 Conflictinstead of a raw database error. An organization ID that is not a UUID, and similar invalid requests, return a 4xx instead of500. (2.77.3) - Monad Logs input shows its synthetic-data setting. (2.77.3)
- Monad MCP —
create_pipelineasks before a new pipeline starts running — omittingenabledcreated a pipeline that was enabled and ingesting, even though the tool documented a disabled default. The tool now declines the call and asks you to choose enabled or disabled, recommending disabled until secrets are filled in and tested. (2.77.3) (For teams using the Monad MCP integration.) - Egress counts no longer include retried records — records routed to a node's retry queue were counted as sent, which overstated records and bytes published for failing or retrying nodes. (2.77.3)
Breaking changes
- Anthropic Chat Messages input: User IDs and Project IDs filters removed — Anthropic deprecated the per-user listing that these filters depend on. The input now keeps its export current from Anthropic's organization-wide listing. Impact: an existing input with User IDs or Project IDs set now collects every chat in the organization, including chats whose creator has left. A chat that fails on every attempt now stops the input with an Erroring status instead of being skipped. As a result, chats updated after they were created, and chats that failed once, are no longer missed. Action: if you relied on the filters, filter the records downstream with an edge condition or a transform. Docs → (2.77.3)
Need help?
See the Monad docs for setup guides and reference, or reach out to Monad Customer Support at support@monad.com or via your dedicated Slack customer channel.
Self-hosted deployments upgrade to a 2.77.x release with
helm upgrade. Review the Breaking changes section above before upgrading;
when it says "None," the upgrade is drop-in.
Last modified on