1. Resources
  2. Detection rules
  3. Password Guessing and Spraying Against OCI Sign-In

Password Guessing and Spraying Against OCI Sign-In detection rules

An attacker tries passwords for one user, or a few common passwords across many users, on the sign-in page of an OCI identity domain. The Audit log shows the failed sign-ins, and a lockout shows that guessing got far enough to block a user.

Platform
Oracle Cloud Infrastructure
Severity
Medium
Log source
Oracle Cloud Infrastructure Audit
Last updated
MITRE ATT&CK technique
T1110Brute Force
Rule formats
Sigmaand queries for: SplunkMicrosoft SentinelIBM QRadar
What an attacker needs
The sign-in URL of the tenancy or domain and user names of accounts that sign in with a local password.
Impact
A session as the user whose password was guessed, in the Console and in every application the identity domain signs that user in to.
Response
If a sign-in succeeded, reset that user's password at once, deactivate the user until the owner is verified, and delete credentials created after the sign-in.
Mitigation
Keep the Console sign-on policy at Oracle's security defaults so a second factor is required, and set an account lock threshold above 0.
Identity domain event
sso.authentication.failure
Event type
com.oraclecloud.identityControlPlane.UpdateUserState

How the attack works

Users of an OCI tenancy sign in to the Console and to integrated applications through an identity domain, with a user name and a password and, where the sign-on policy asks for it, a second factor. The sign-in page is reachable from the internet, so anyone who knows or guesses user names can try passwords. The attempts take the usual forms: many passwords for one user, a few common passwords across many users, or user name and password pairs taken from breaches of other services.

Two controls limit this, and both are settings an administrator can weaken. The password policy of an identity domain has an account lock threshold, the number of consecutive unsuccessful sign-in attempts after which the account is locked. Oracle states that a value of 0 means the account is never locked, and that a locked account can be unlocked automatically after a period between 5 minutes and 24 hours. In a tenancy without identity domains a user is blocked after 10 unsuccessful Console sign-ins in a row and stays blocked until an administrator unblocks it. A correct guess is only stopped by the second factor.

The identity domain writes each failed sign-in to the Audit log with the event id sso.authentication.failure in data.additionalDetails.eventId. Oracle's own report query for failed sign-ins reads the user from data.additionalDetails.actorName, the domain from data.additionalDetails.domainDisplayName and the reason from data.message. One failed sign-in is a typing mistake. The two threshold rules on this page count failures per user and distinct users per client address, and a third rule reads com.oraclecloud.identityControlPlane.UpdateUserState, the operation that unblocks a blocked user. No public incident report describes password guessing against OCI at the audit event level. The page follows what Oracle documents about the events and the lockout settings. Collection needs the _Audit log group of every compartment that holds an identity domain, because the sign-in events are recorded in the compartment of the domain.

What an attacker needs

  • The sign-in URL of the tenancy or of an identity domain, and one or more user names.
  • Accounts that sign in with a local password. Users who authenticate at a federated identity provider enter their password there, and a wrong password at the identity provider does not produce these events.

Impact

A session as the user whose password was guessed, in the Console and in every application the domain signs that user in to. With a second factor required by the sign-on policy, the password alone does not complete the sign-in. Without it, one correct guess is a session, from which the attacker can create API keys and further users. Repeated guessing also locks legitimate users out when the account lock threshold is reached.

Detection rules

Failed Sign-Ins for Many OCI Identity Domain Users From One Client Address

Counts, per client address (data.additionalDetails.clientIp), the distinct users (data.additionalDetails.actorName) whose sign-in to an OCI identity domain failed (event id sso.authentication.failure). Ten or more users from one address within 30 minutes fit password spraying: a few common passwords tried across many accounts, which stays under the account lock threshold of each user. The count and the window are starting points to tune. The grouping field needs a check before you rely on the rule. Oracle documents clientIp as an attribute of identity domain audit events (the IP address of the client application making the request), and Oracle's report queries show that such attributes sit in data.additionalDetails in the Audit log, but Oracle publishes no Audit sample that shows clientIp there. Confirm the field on one failed sign-in of your own. If it is absent, the rule never fires and data.identity.ipAddress is the field to test in its place.

Fires when one data.additionalDetails.clientIp has at least 10 different data.additionalDetails.actorName values within 30 minutes.

Threshold ruleLevelMediumStatusExperimentalLog sourceOracle Cloud Infrastructure Audit
title: OCI Identity Domain Sign-In Failed (Password Spraying Base Event)
id: f84586d7-0e6c-436d-9bf5-45a6346b4f34
name: oci_identity_domain_sign_in_failed_per_address
status: experimental
description: >-
  Base event of the threshold rule in this file. An Audit log event whose
  data.additionalDetails.eventId is sso.authentication.failure, the identity
  domain event id that Oracle lists as User Logins Failure.
logsource:
  product: oci
  service: audit
detection:
  selection:
    data.additionalDetails.eventId: sso.authentication.failure
  condition: selection
level: low
---
title: Failed Sign-Ins for Many OCI Identity Domain Users From One Client Address
id: ba1dcfbb-6614-4a74-9c47-9689008e36b1
status: experimental
description: >-
  Counts, per client address (data.additionalDetails.clientIp), the distinct
  users (data.additionalDetails.actorName) whose sign-in to an OCI identity
  domain failed (event id sso.authentication.failure). Ten or more users from
  one address within 30 minutes fit password spraying: a few common passwords
  tried across many accounts, which stays under the account lock threshold of
  each user. The count and the window are starting points to tune. The
  grouping field needs a check before you rely on the rule. Oracle documents
  clientIp as an attribute of identity domain audit events (the IP address of
  the client application making the request), and Oracle's report queries
  show that such attributes sit in data.additionalDetails in the Audit log,
  but Oracle publishes no Audit sample that shows clientIp there. Confirm the
  field on one failed sign-in of your own. If it is absent, the rule never
  fires and data.identity.ipAddress is the field to test in its place.
references:
  - https://docs.oracle.com/en-us/iaas/Content/Identity/api-getstarted/usingauditeventapis.htm
  - https://docs.oracle.com/en/cloud/paas/iam-domains-rest-api/op-admin-v1-auditevents-id-get.html
  - https://docs.oracle.com/en/learn/generating-iam-reports-from-oci-audit/index.html
  - https://attack.mitre.org/techniques/T1110/003/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1110
correlation:
  type: value_count
  rules:
    - oci_identity_domain_sign_in_failed_per_address
  group-by:
    - data.additionalDetails.clientIp
  timespan: 30m
  condition:
    field: data.additionalDetails.actorName
    gte: 10
falsepositives:
  - Many people behind one office, VPN or proxy address who mistype their password in the same period, for example after a forced password change. Exclude your own egress addresses or give them a higher count.
  - A shared application or kiosk that signs many users in from one address.
  - The values of 10 users and 30 minutes are starting points, not tested limits. A spraying source that changes its address for every attempt is not seen by a count per address.
level: medium

False positives

  • Many people behind one office, VPN or proxy address who mistype their password in the same period, for example after a forced password change. Exclude your own egress addresses or give them a higher count.
  • A shared application or kiosk that signs many users in from one address.
  • The values of 10 users and 30 minutes are starting points, not tested limits. A spraying source that changes its address for every attempt is not seen by a count per address.

Many Failed Sign-Ins for One OCI Identity Domain User

Counts, per user (data.additionalDetails.actorName), the failed sign-ins to an OCI identity domain (event id sso.authentication.failure). Ten or more within 15 minutes are more than a forgotten password and fit password guessing against that user. Oracle's report query for failed sign-ins reads the user from the same field, the domain from data.additionalDetails.domainDisplayName and the reason from data.message. The count and the window are starting points to tune. Set the count at or below the account lock threshold of your password policy: Oracle does not say whether attempts against a locked account still produce this event, so a count above the threshold may never be reached. Oracle also does not document what actorName holds for a user name that does not exist. Users who authenticate at a federated identity provider do not produce this event for a wrong password.

Fires when one data.additionalDetails.actorName has at least 10 matching events within 15 minutes.

Threshold ruleLevelMediumStatusExperimentalLog sourceOracle Cloud Infrastructure Audit
title: OCI Identity Domain Sign-In Failed (Password Guessing Base Event)
id: 271b1713-b260-4ce8-a008-32810ce42142
name: oci_identity_domain_sign_in_failed_per_user
status: experimental
description: >-
  Base event of the threshold rule in this file. An Audit log event whose
  data.additionalDetails.eventId is sso.authentication.failure, the identity
  domain event id that Oracle lists as User Logins Failure.
logsource:
  product: oci
  service: audit
detection:
  selection:
    data.additionalDetails.eventId: sso.authentication.failure
  condition: selection
level: low
---
title: Many Failed Sign-Ins for One OCI Identity Domain User
id: 2f1c628c-df73-40a9-85aa-e0b6aa8a919e
status: experimental
description: >-
  Counts, per user (data.additionalDetails.actorName), the failed sign-ins to
  an OCI identity domain (event id sso.authentication.failure). Ten or more
  within 15 minutes are more than a forgotten password and fit password
  guessing against that user. Oracle's report query for failed sign-ins reads
  the user from the same field, the domain from
  data.additionalDetails.domainDisplayName and the reason from data.message.
  The count and the window are starting points to tune. Set the count at or
  below the account lock threshold of your password policy: Oracle does not
  say whether attempts against a locked account still produce this event, so
  a count above the threshold may never be reached. Oracle also does not
  document what actorName holds for a user name that does not exist. Users
  who authenticate at a federated identity provider do not produce this event
  for a wrong password.
references:
  - https://docs.oracle.com/en-us/iaas/Content/Identity/api-getstarted/usingauditeventapis.htm
  - https://docs.oracle.com/en/learn/generating-iam-reports-from-oci-audit/index.html
  - https://docs.oracle.com/en-us/iaas/Content/Identity/passwordpolicies/Managing-Password-Policies_set-password-policies-your-identity-domain.htm
  - https://attack.mitre.org/techniques/T1110/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1110
correlation:
  type: event_count
  rules:
    - oci_identity_domain_sign_in_failed_per_user
  group-by:
    - data.additionalDetails.actorName
  timespan: 15m
  condition:
    gte: 10
falsepositives:
  - A password manager, mobile app or script that keeps submitting an old password for one user after a password change.
  - A user who repeatedly mistypes a new password. The reason in data.message and a known source address help to tell this apart.
  - The values of 10 failures and 15 minutes are starting points, not tested limits. Tenancies with a low account lock threshold need a lower count.
level: medium

False positives

  • A password manager, mobile app or script that keeps submitting an old password for one user after a password change.
  • A user who repeatedly mistypes a new password. The reason in data.message and a known source address help to tell this apart.
  • The values of 10 failures and 15 minutes are starting points, not tested limits. Tenancies with a low account lock threshold need a lower count.

OCI IAM User Unblocked After Failed Sign-In Attempts

Detects the IAM control plane operation UpdateUserState. Oracle documents that a user who unsuccessfully tries to sign in to the Console 10 times in a row is blocked from further attempts, and that an administrator unblocks the user in the Console or with UpdateUserState. The operation takes one setting, blocked, and Oracle states that only the value false is supported, for changing the state to unblocked. Every event of this type therefore means that a user had been blocked by failed sign-ins and was let in again. That is evidence of guessing against the user named in data.resourceName, and the unblock itself deserves a look when the actor (data.additionalDetails.actorName) is not the help desk. Oracle describes the blocking for tenancies without identity domains. In an identity domain the lock comes from the password policy and can end automatically, so the absence of this event does not mean that nobody was locked.

LevelLowStatusExperimentalLog sourceOracle Cloud Infrastructure Audit
title: OCI IAM User Unblocked After Failed Sign-In Attempts
id: 05f496a5-37f9-476d-aff1-a877503ddecb
status: experimental
description: >-
  Detects the IAM control plane operation UpdateUserState. Oracle documents
  that a user who unsuccessfully tries to sign in to the Console 10 times in
  a row is blocked from further attempts, and that an administrator unblocks
  the user in the Console or with UpdateUserState. The operation takes one
  setting, blocked, and Oracle states that only the value false is supported,
  for changing the state to unblocked. Every event of this type therefore
  means that a user had been blocked by failed sign-ins and was let in again.
  That is evidence of guessing against the user named in data.resourceName,
  and the unblock itself deserves a look when the actor
  (data.additionalDetails.actorName) is not the help desk. Oracle describes
  the blocking for tenancies without identity domains. In an identity domain
  the lock comes from the password policy and can end automatically, so the
  absence of this event does not mean that nobody was locked.
references:
  - https://docs.oracle.com/en-us/iaas/Content/Events/Reference/eventsproducers.htm
  - https://docs.oracle.com/en-us/iaas/Content/Identity/Tasks/managingusers.htm
  - https://docs.oracle.com/en-us/iaas/tools/oci-cli/latest/oci_cli_docs/cmdref/iam/user/update-user-state.html
  - https://attack.mitre.org/techniques/T1110/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1110
logsource:
  product: oci
  service: audit
detection:
  selection:
    eventType: com.oraclecloud.identityControlPlane.UpdateUserState
  condition: selection
falsepositives:
  - Users who forgot their password and asked the help desk to unblock them. Confirm with the user that the ten failed attempts were their own, and review unblocks of administrators whoever made them.
level: low

False positives

  • Users who forgot their password and asked the help desk to unblock them. Confirm with the user that the ten failed attempts were their own, and review unblocks of administrators whoever made them.

Detection logic

Patterns that need more than one event, a baseline or knowledge of your own environment. Each is a hypothesis to build and tune in your SIEM, not a ready rule.

Run of failed sign-ins followed by a successful sign-in for the same user

Detection hypothesisSequence of eventsNeeds a baseline

Guessing matters when it works. A user with several failed sign-ins and then a successful one within a short time may have mistyped a password, but the same pattern is what a correct guess looks like. It is the point where the count of failures turns into an account to investigate.

  1. The Audit log records several events with sso.authentication.failure in data.additionalDetails.eventId for one data.additionalDetails.actorName. Five within 15 minutes is a starting point.
  2. Within 30 minutes of the last failure the Audit log records sso.session.create.success for the same data.additionalDetails.actorName in the same data.additionalDetails.domainDisplayName.
  3. Compare the source of the failures with the source of the success, and check whether the address is one the user has signed in from before.

Eventssso.authentication.failuresso.session.create.success

LimitsA user who forgot a password and then remembered it produces the same sequence, usually from a known address. The counts and windows are starting points. The events do not say whether a second factor was asked for, so a success after failures may still have been stopped later in the sign-in.

Triage

  • Read who was targeted (data.additionalDetails.actorName) and in which domain (data.additionalDetails.domainDisplayName). One user with many failures is guessing or a stale saved password. Many users with a few failures each is spraying.
  • Read data.message for the reason Oracle records with the failure.
  • Identify where the attempts came from. Oracle lists clientIp among the attributes of an identity domain audit event, and the Audit log event has data.identity.ipAddress. An address of your own network with one affected user is usually a forgotten password.
  • Search for sso.session.create.success for the same users during and after the failures. A success after a run of failures is the case that needs action.
  • Check whether the targeted users are members of Administrators or hold an administrator role, and whether any of them was locked or unblocked in the same period.
  • If a sign-in succeeded, review what that user did afterwards, with priority on new API keys, new users, group changes and policy changes.

Response

Once the attack is confirmed:

  1. If a sign-in succeeded, reset the password of that user at once and deactivate the user until the owner is verified. Oracle notes that a deactivated user keeps the current session with limited capabilities until it ends, so watch that session.
  2. Delete API keys, auth tokens and other credentials that were created for the user after the sign-in, and undo the changes the session made.
  3. Enroll a second factor for every targeted user before the account is used again, and confirm that the sign-on policy requires it.
  4. Block the source addresses where the organization controls a network perimeter or a network source restriction, knowing that a spraying source changes addresses.
  5. Keep the threshold rules on a lower count for the targeted users for the following weeks, because the attacker now knows that they exist.

Mitigation

  • Keep the Console sign-on policy at Oracle's security defaults, which require a second factor, and prefer phishing-resistant factors for administrators.
  • Set an account lock threshold above 0 in every password policy, and decide deliberately whether locked accounts unlock automatically.
  • Restrict Console sign-in to known networks where the organization allows it. Oracle documents network source restrictions for this in the authentication settings.
  • Sign people in through a federated identity provider and keep local passwords for break-glass accounts only.
  • Collect the Audit log of every compartment that holds an identity domain, not only the root compartment.

Frequently asked questions

Which Audit log event shows a failed sign-in to an OCI identity domain?

The identity domain event id `sso.authentication.failure`, which Oracle lists as "User Logins Failure". It appears in the Audit log in `data.additionalDetails.eventId`. Oracle's report query for failed sign-ins reads the user from `data.additionalDetails.actorName` and the reason from `data.message`. A successful sign-in has the id `sso.session.create.success`.

Does OCI lock an account after repeated failed sign-ins?

It depends on the settings. In an identity domain the password policy has an account lock threshold, and Oracle states that a value of 0 means the account is never locked. A locked account can unlock automatically after 5 minutes to 24 hours if that option is enabled. In a tenancy without identity domains a user is blocked after 10 failed Console sign-ins in a row until an administrator unblocks it.

Why does the spraying rule ask me to confirm a field first?

The rule groups failures by `data.additionalDetails.clientIp`. Oracle documents clientIp as an attribute of identity domain audit events, the address of the client making the request, and its report queries show that such attributes sit in `data.additionalDetails` in the Audit log. Oracle publishes no Audit sample that shows this one. Check one failed sign-in of your own before you rely on the rule.

Do these rules see wrong passwords of federated users?

No. A user who authenticates at a federated identity provider enters the password there, so the wrong password is rejected by the identity provider and logged in its own sign-in log. The identity domain only sees the assertion that follows a successful authentication. Watch the identity provider's log for those accounts.

References

We respect your privacy

We use cookies to keep this site secure and working properly. With your permission, we also use optional cookies to understand usage and improve the experience. Cookie Policy

You can change your choice at any time.