Skip to content

Error codes

When CloudRepo refuses a request, the response carries a code that says why. This page lists the codes documented so far, with what causes each one and how to fix it. Search the page for the code you got, or follow a link that ends in # and the code.

Two hosts answer with these codes:

  • Your repository host, your-org.mycloudrepo.io, which your build tools use to publish and fetch packages.
  • The CloudRepo API, api.cloudrepo.io, which the admin portal uses and which you can call yourself.

Where the code is depends on the response:

  • Most JSON responses carry it in the error field, for example {"error": "token_revoked"}.
  • When your repository host refuses a package request with 403, the error field holds a sentence and the code is in the reason_code field, for example {"error": "You do not have permissions to access the repository.", "reason_code": "scope_mismatch_repo"}.
  • A few responses carry it in a code field; their entries say so.
  • Docker answers in the registry’s own format, with an upper-case code in errors[].code. Those codes are at the end of this page.

The codes that single sign-on (SSO) answers with are on Error codes for SSO.

Used on this page only, never saved. Never enter a token or password here.

Many build tools print only the status, such as 403 Forbidden, and not the code. To read the code, repeat the request with curl and show the response. Keep your credential in ~/.netrc:

~/.netrc
machine your-org.mycloudrepo.io
login you@example.com
password YOUR_REPOSITORY_TOKEN
Terminal window
curl --include --netrc https://your-org.mycloudrepo.io/repositories/your-repo/

Status 400, from the CloudRepo API, when you create a repository token.

Why: the token’s name is empty, is longer than 256 characters, or contains a control character or a character that reverses text direction.

Fix: give the token a short, plain name, such as the name of the build that uses it.

Status 400, from the CloudRepo API, when you create a repository token.

Why: the request named no scope, or a scope other than read and write.

Fix: ask for read, write, or both.

Status 400, from the CloudRepo API, when you create a repository token.

Why: the request asked for a token format other than generic and npm.

Fix: ask for generic, or npm for a token that npm uses.

Status 400, from the CloudRepo API, when you create a repository token.

Why: the request named no repository.

Fix: name at least one repository the token may reach.

Status 400, from the CloudRepo API, when you create a repository token or change an access key’s expiry.

Why: the request did not say when the token expires. Choosing no expiry is allowed, but it has to be asked for.

Fix: send an expiry date, or null for a token that never expires.

Status 400, from the CloudRepo API, when you create a repository token, rotate one with a new expiry, or create an access key or change its expiry.

Why: the expiry is not a date CloudRepo can store: it is not a whole number of milliseconds, or it is after the end of the year 9999.

Fix: send the expiry as milliseconds since 1970 (UTC), or null for a token that never expires.

Status 400, from the CloudRepo API, when you create a repository token, rotate one with a new expiry, or create an access key or change its expiry.

Why: the expiry you chose has already passed.

Fix: choose a date in the future.

Status 400, from the CloudRepo API, when you create or rotate a service key.

What you see, for an organization token:

A service key must have an expiry date no more than 365 days after it is created or rotated. Only a token that belongs to a person can have a later date, or no expiry.

For any other service key, such as one that belongs to a single repository, the message says 90 days.

Why: a service key belongs to your organization, not to a person, so it must end. An organization token can end at most 1 year after you create or rotate it, and any other service key at most 90 days after.

Fix: choose an earlier date. Before a key ends, make the next one and put it in your CI.

Status 400, from the CloudRepo API, when you create a repository token.

Why: the request’s backing.kind is not one CloudRepo knows. A service key belongs to the organization (org) or to one repository (repo).

Fix: send org or repo for a service key, or leave backing out for a token of your own.

Status 400, from the CloudRepo API, when you create a service key that belongs to a repository.

Why: the request did not say which repository the key belongs to.

Fix: send that repository’s ID as backing.id.

Status 400, from the CloudRepo API, when you create a service key.

Why: service keys are created only in the generic format.

Fix: ask for the generic format.

Status 400, from the CloudRepo API, when you create a service key that belongs to a repository.

Why: a key that belongs to one repository can reach only that repository.

Fix: list only the key’s own repository in the request.

Status 400, from the CloudRepo API, when you create or rotate a repository token.

Why: creating or rotating a token needs a fresh confirmation of who you are, and the request carried none.

Fix: in the portal, confirm your identity when it asks, then try again.

Status 400, from the CloudRepo API, when you rotate a repository token or an access key.

What you see:

This token is already being rotated and stops working at its scheduled time. A new rotation cannot keep it working longer: choose a shorter overlap, or 0 to end it now.

For an access key, the message begins This access key instead of This token.

Why: the token is already being rotated, and a second rotation asked to keep it working past the end the first one set.

Fix: choose a shorter overlap, or 0 to stop the old token now.

Status 400, from the CloudRepo API, when you list repository tokens for a member.

Why: the request named a member, but the name was blank.

Fix: name the member, or leave the parameter out to list your own tokens.

Status 400, from the CloudRepo API, when you rotate an access key or a repository token.

Why: the overlap, how long the old key keeps working after the rotation, is outside 0 to 168 hours.

Fix: choose an overlap from 0, which stops the old key now, to 168 hours, which is one week.

Status 400, from the CloudRepo API, when you rotate a repository token.

Why: the overlap, how long the old token keeps working after the rotation, is longer than the time the old token has left before it expires. With no overlap chosen, the overlap is 24 hours.

Fix: choose an overlap no longer than the hours the old token has left, or 0 to stop the old token now.

Status 400, from the CloudRepo API, when you rotate a repository token.

Why: CloudRepo found the token when the rotation began, but not when it went to replace it, for example because the token was deleted in between.

Fix: reload your list of tokens. If you still need one, create a new token.

Status 400, from the CloudRepo API, when the member list, GET /api/admin/sso/users, is asked for with a cursor it cannot use. The response carries the code in the error field and no sentence.

Why: The cursor you sent cannot be used. It was empty or only spaces, it was sent more than once, it was longer than 1015 bytes, or it held a control character.

Fix: ask for the first page with no cursor. For each next page, send the next-cursor the page before gave you, once and unchanged, as cursor. Other lists in the CloudRepo API answer invalid_cursor too, when the value you send to continue them cannot be used.

Status 400, from the CloudRepo API, when you add a repository to a group. The code is in the response’s code field.

Why: the group is public, and a public group can hold only public repositories. The repository you added is private.

Fix: make the repository public, or make the group private, then add the repository again.

Status 400, from the CloudRepo API, when you turn on Public Access for a group. The code is in the response’s code field.

Why: a public group can hold only public repositories, and this group holds private ones. The message names them.

Fix: make those repositories public, or remove them from the group, then make the group public.

Status 400, from the CloudRepo API, when you create an npm proxy for your own registry. The code is in the code field.

Why: the request did not list the scopes your registry owns. Through a group, CloudRepo never serves a package in those scopes from npmjs.

Fix: list your registry’s scopes in npmScopes, such as ["@acme"], or set it to "none".

Status 400, from the CloudRepo API, when you create an npm proxy for your own registry. The code is in the code field.

Why: a scope in npmScopes is not an @ and a lower-case name, is listed twice, or the list holds more than 50.

Fix: list each scope once, in the form @acme, with 50 at most.

Status 400, from the CloudRepo API, when you add a repository to a Docker group. The code is in the code field.

What you see:

CloudRepo can't add your own Docker registry to a group yet. Use it directly as its own repository.

Why: the repository is a Docker proxy for your own registry, and a Docker group cannot hold one.

Fix: pull from that proxy directly, not through the group.

Status 400, from the CloudRepo API, when you ask to be an Admin.

What you see:

from must be one of access-log, audit-log, billing, software-distribution, users, webhooks.

Why: the request’s from field is missing, or it does not name one of those six pages. The portal’s Request Admin button sends from for you, so you meet this when you call the API yourself.

Fix: send from as the page you are asking from, spelled as one of the six names.

Status 400, from the CloudRepo API, when you subscribe to a plan or change your card.

What you see:

We didn't receive your card details. Refresh the page and try again.

Why: the request carried no card token, so the payment form did not send your card details. The request is refused before it counts as one of your organization’s card attempts for the day.

Fix: refresh the page and enter the card again. If you call the API yourself, send the card token the payment form makes in nonce.

Status 400, from the CloudRepo API, when you finish an npm login in your browser. The code is in the code field.

What you see:

npm login works on a Local repository. For a Proxy or a group, use a repository token.

Why: the repository you signed in to is a proxy or a group. A token made by npm login is not accepted on those (see npm_token_not_supported_on_proxy), so CloudRepo makes none.

Fix: use a repository token for that repository. The steps are in In CI: a token in an environment variable.

Status 400, from your repository host, when you finish an npm login in your browser for an npm group. The code is in the code field.

What you see:

npm login is not supported for group repositories. Use a repository token that names the group and its members, in .npmrc as _authToken, or Basic auth as _auth. See https://docs.cloudrepo.io/ for setup.

Why: npm login does not work on a group repository.

Fix: use a repository token that names the group and its members. Put it in .npmrc as _authToken, or send it as Basic auth in _auth. The steps are in npm groups.

Status 401 when you create or rotate a repository token, or finish an npm login. Status 403 for other sensitive actions, such as managing a subscriber’s access keys or setting a proxy repository’s upstream credentials. From the CloudRepo API.

Why: the action needs a fresh confirmation of who you are. The request carried none, or one that has expired or was given for a different action. Creating or rotating a token with none answers missing_confirmation_token instead.

Fix: in the portal, confirm your identity again when it asks, then repeat the action.

What you see, from Maven, Gradle, npm or pip (status 401):

Password authentication is disabled because MFA is required for this organization. Use a repository token instead.

From Docker (status 401, with the registry code UNAUTHORIZED), the same sentence followed by where to create a token.

Why: your organization requires MFA, and the build signed in with a member’s password.

Fix: use a repository token instead of the password. A pipeline should use a service key, which belongs to the organization rather than to a person.

Status 401, from your repository host.

What you see:

This organization accepts repository tokens only, not passwords. Create a token and use it in place of your password.

Why: your organization turned password builds off, and the build signed in with a password.

Fix: create a repository token and use it where the build had the password. The response’s tokenPageUrl opens the page where you create one, and docsUrl shows where the token goes in each client.

Status 401, from the CloudRepo API, when you choose a new password from a reset link.

What you see:

This link was replaced by a newer one, so your password wasn't changed. Use the newest link in your email, or ask an owner or Admin to send another.

Why: an owner or an Admin sent you a newer reset link after you used yours and before your new password was saved. The older link stopped working, and your password was not changed.

Fix: open the newest link in your email. If you cannot find it, ask an owner or an Admin to send another.

Status 402, from your repository host, when you upload. The code is in the reason_code field.

What you see:

Your CloudRepo trial has ended, so uploads to this organization are paused. Pick a plan to resume uploads: <your Billing page>

Why: the organization’s trial has ended, so uploads are paused.

Fix: pick a plan on the Billing page the message names. Uploads work again once you have one.

Status 403, from your repository host, when you read through a group.

What you see:

You can't read a repository in this group, so CloudRepo won't serve this file from the group's other repositories.

Why: your credential cannot read one of the group’s repositories. CloudRepo then serves the file from none of them, so a build never gets a public package in place of your team’s private one.

Fix: use a token that can read every repository in the group, or ask an owner or Admin to give you read access to each one.

Status 403, from the CloudRepo API, when you create a repository token or finish an npm login.

Why: the token would reach a repository that you cannot access yourself.

Fix: choose only repositories you can access, or ask your organization’s owner for access first.

When an Admin rotates an organization token, what you see:

This token names a repository that no longer exists, so only your organization's owner can rotate it. You can create a new organization token instead.

Why: the token names a repository that no longer exists, so only your organization’s owner can rotate it.

Fix: ask the owner to rotate it, or create a new organization token.

Status 403, from the CloudRepo API, when you create a service key.

What you see:

Only your organization's owner or an Admin can create an organization token.

Why: only the organization’s owner or an Admin can create a key that belongs to the organization.

Fix: ask your organization’s owner or an Admin to create the key.

Status 403, from the CloudRepo API, in each of these cases:

When Why Fix
You create a token with write Writing needs both publish and delete access, and you do not hold both in every repository the token would reach. Create a read token, or ask your organization’s owner for publish and delete access to those repositories first.
You create a user, invite one, or change a member’s permissions You asked for a permission you do not hold yourself. Grant only permissions you hold, or ask your organization’s owner to make the change.
You are a read-only member and you give more access to a member who can write, or add or invite one A read-only member can give more access to, add and invite only members who cannot write. Turning a switch off, turning on Read only, removing a member and resetting a password are not refused for this reason. Ask a member who can write, or your organization’s owner.
You reset a member’s password That member has an administrator permission that you do not have. Your own access to repositories does not matter. Ask your organization’s owner to reset it.

Status 403, from the CloudRepo API, when you list another member’s repository tokens.

Why: only a member who can manage the organization’s owner account can list another member’s tokens.

Fix: list your own tokens, or ask your organization’s owner.

Status 403, from your repository host. The code is in the response’s reason_code field, and the error field holds one of these sentences:

You do not have permissions to access the repository. [2242]
You do not have the required permissions [read] for this repository.
This token can download but not publish. Create one with Download and publish access.
This token doesn't include this repository. On the Tokens page, make a token that includes it.
This CI token's trust binding doesn't include this repository. On the Trusted publishing page, revoke that binding and register one that includes it.

The second names the action that was refused. The third means your token was made for download only, and a new token with Download and publish access can publish or delete here. The fourth means your token was made for other repositories; a token that includes this one works here. The fifth is the fourth for a CI token from trusted publishing: its trust binding names other repositories. A binding can’t be edited, and one CI job can’t have two, so revoke the old binding and register one that includes this repository.

Why: the credential is valid, but it does not allow this action on this repository. It does not reach the repository, or it can only read and you published or deleted. An access key can only read, and only the repositories it was given. A token in the npm format works only on npm repositories.

Fix: use a repository token that reaches this repository, with write scope to publish or delete. If you sign in with your own account, ask your organization’s owner for access to the repository. With an access key, ask the organization that gave it to you for access to this repository; an access key can never publish or delete. For a repository that is not npm, replace a token in the npm format with a repository token created in the portal.

Status 403, from your repository host, when you publish or delete with a repository token. The code is in the response’s reason_code field.

What you see:

Your credential's backing authority no longer permits this action; the credential was clamped at use time. [7301]

Why: the token was created with write, but the member it belongs to can no longer publish or delete in this repository.

Fix: ask your organization’s owner to give that member write access to the repository again, or use a token that belongs to a member who can write there.

Status 403, from your repository host. The code is in the response’s code field.

Why: npm sent a token in the npm format to a proxy repository or a group repository. Neither takes npm-format tokens.

Fix: use a repository token in the generic format. For a group, the token must name the group and its members.

Status 403, from the CloudRepo API, when you sign in to the portal.

Why: your multi-factor authentication must be reset before you can sign in.

Fix: if the portal says your multi-factor authentication must be reset, ask your organization’s owner to reset it. See Resetting a member’s MFA.

Status 403, from the CloudRepo API, when you sign in to the portal.

Why: your second factor can’t be used on this site, and you have no backup codes left.

Fix: if the portal says your second factor can’t be used here and you have no backup codes left, ask your organization’s owner to reset your MFA, or contact support. See If someone loses their authenticator.

Status 403, from the CloudRepo API, when a member signs in to the portal with a password.

What you see: the code, with two flags and no sentence:

{ "error": "mfa_enrol_unavailable_legacy_okta", "org_mfa_required": true, "mfa_reset_pending": false }

Why: your organization is on the legacy Okta integration, you have no second factor, and sign-in is asking you to set one up. That happens when the organization requires MFA (org_mfa_required is true), or when your factor was reset and not yet replaced (mfa_reset_pending is true). On the legacy Okta integration only the owner can set up MFA, so you cannot finish here. No session was made.

Fix: ask your organization’s owner. When org_mfa_required is true and mfa_reset_pending is false, the owner can turn the requirement off, which lets you in. In either case the owner can contact support about moving the organization off the legacy Okta integration.

Status 403, from your repository host, when a write to a repository is sent by a web page through your browser.

What you see: A write to a repository sent by a web page is refused.

Why: a web page tried to write to the repository through your signed-in browser, and CloudRepo refuses writes that come from another site.

Fix: upload with your package tool or CI as usual. If the write came from a page you did not expect, close it.

Status 403, from the CloudRepo API, when you act on an Admin’s account or the owner’s.

What you see:

Only the owner can change an Admin or the owner.

Why: only the owner acts on an Admin, and an Admin cannot act on the owner. An Admin who changes another Admin’s or the owner’s permissions, resets their password, rotates their credentials, removes them or lists their repository tokens gets this answer. The same code answers an invitation sent to an Admin’s email address, from the owner as well, with its own sentence: This person is an Admin. Change their access on the Users page.

Fix: ask the owner to make the change. To change an Admin’s access, use the Users page as the owner.

Status 403, from the CloudRepo API, when you create a proxy repository for your own registry.

What you see:

Custom upstreams are in early access. Contact us to turn them on for your organization.

Why: a proxy whose upstream is your own registry, not one of the registries CloudRepo lists, is in early access. An organization on a paid plan is admitted, and so is a trial that we turn on by hand. This one is neither.

Fix: contact us and we will turn it on for your organization. Once your organization is on a paid plan, this refusal no longer applies.

Status 403, from the CloudRepo API, when you invite someone.

Why: the invitation includes a Manage permission. An invitation carries repository access only, because the person cannot yet have the two-factor authentication a Manage permission needs. The message says so, and says to give the permission on the Users page after the person has joined and turned on two-factor authentication.

Fix: invite the person with repository access. After they join and turn on two-factor authentication, give them the Manage permission on the Users page.

Status 404, from the CloudRepo API, when you list a member’s repository tokens.

Why: no member of your organization has that name.

Fix: check the member’s email address in the portal’s user list.

Status 404, from the CloudRepo API, when you manage a subscriber’s access key.

Why: there is no access key with that ID for this subscriber. A key that belongs to a different subscriber gets the same answer.

Fix: check the key’s ID in the subscriber’s list of access keys.

Status 404, from the CloudRepo API, when the owner presses Not now on a request to be an Admin.

What you see:

This member has no Admin request waiting.

Why: no request from that member is waiting in your organization. The member has not asked in the last 7 days, or a Not now already answered the request.

Fix: reload the Users page to see the requests that are waiting.

Status 405, from your repository host or the CloudRepo API. The code is in the response’s code field.

What you see:

Writes are not permitted to Proxy Repositories.

Why: you published or uploaded to a proxy repository. A proxy repository only holds what it fetches from its upstream.

Fix: publish to a hosted repository instead.

Status 408, from the CloudRepo API.

What you see:

The request body did not arrive.

Why: the request said it had a body, but the body never reached CloudRepo.

Fix: send the request again.

Status 409, from the CloudRepo API, when you rotate a repository token.

What you see:

This token has been revoked, so it cannot be rotated. Create a new token instead.

Why: the token was revoked.

Fix: create a new token.

Status 409, from the CloudRepo API, when you rotate a repository token.

Why: the token does not record whose token it is, so CloudRepo cannot rotate it. Tokens created before CloudRepo recorded that are like this.

Fix: create a new token instead.

Status 409, from the CloudRepo API, when you rotate an access key or change its expiry.

What you see:

This access key has been revoked, so it cannot be rotated. Create a new access key instead.

Or, when you change its expiry:

This access key is being rotated or has been revoked, so its expiry cannot be changed.

Why: the key was replaced when it was rotated. Once its overlap has ended it cannot be rotated again, and once a rotation has begun its expiry cannot change.

Fix: use the new key the rotation created, and rotate it or change its expiry instead.

Status 409, from the CloudRepo API, when you change an access key’s expiry.

What you see:

This access key changed while you edited it. Reload it and try again.

Why: the key changed after you opened it.

Fix: reload the page and make your change again.

Status 409, from your repository host, when you unpublish or delete an npm package or version.

Why: the package changed after npm read it, for example because another publish or unpublish finished first.

Fix: run the command again, so npm reads the package’s current state first.

Status 409, from the CloudRepo API, when you accept an invitation.

Why: your organization already has an account at this email address. The invitation is used up.

Fix: sign in with the account you already have. If you cannot, ask your organization’s owner.

What you see, when you turn off MFA for your own account:

Your organization requires multi-factor authentication, so it cannot be turned off for your account.

Why: while the organization requires MFA, nobody can remove their own second factor, the owner included.

Fix: keep your second factor. If you lost your authenticator, sign in with one of your backup codes, then move MFA to a new device: the owner uses Reset authenticator app on their own account’s security page, and a member asks the owner to use Reset MFA for them. The next sign-in sets up the new device. An owner who wants MFA off for their own account first stops requiring MFA for the organization.

Status 409, from the CloudRepo API, when you ask to be an Admin.

What you see:

You are the owner of this organization.

Why: you are the owner of the organization, and the owner does not ask to be an Admin.

Fix: nothing to do. To make a member an Admin, use Make Admin on the Users page.

Status 409, from the CloudRepo API, when you ask to be an Admin.

What you see:

You are already an Admin.

Why: your account is already an Admin in this organization, so the request has nothing to add.

Fix: nothing to do.

Status 409, from the CloudRepo API, when you ask to be an Admin.

What you see:

Your account is not active in this organization.

Why: your account no longer counts as an active member of this organization. A member who has left or is leaving is in this state, and so is one whose account is suspended.

Fix: ask your organization’s owner about your account.

Status 409, from the CloudRepo API. Three actions answer with it, each about two-factor authentication and the Admin role.

When you ask to be an Admin, what you see:

Turn on two-factor authentication first. Every Admin uses it.

Why: you sign in with a password and have no second factor on your account.

Fix: set up two-factor authentication on your account, then ask again.

When you give a member the Admin role or a Manage permission, what you see:

This member needs to turn on two-factor authentication before they can be made an Admin.

For a Manage permission the sentence says This member must turn on two-factor authentication before they can be given this permission.

Why: the member signs in with a password and has no second factor, so CloudRepo does not give them the role or the permission.

Fix: ask the member to turn on two-factor authentication, then give it again.

When an Admin, or a member who holds a Manage permission, turns off their own two-factor authentication, what you see:

An Admin must keep two-factor authentication on. Ask the owner to remove your Admin role first.

A member with a Manage permission reads Two-factor authentication must stay on while you have a Manage permission. Ask an Admin or the owner to remove your Manage permissions first.

Why: an Admin, and a member with a Manage permission, must keep two-factor authentication on while they hold it.

Fix: leave two-factor authentication on, or ask the owner (or, for a Manage permission, an Admin or the owner) to remove the role or the permission first. Then you can turn it off.

Status 409, from the CloudRepo API, when you turn on a permission for a user.

What you see:

This user shares the organization owner's sign-in address, so it has repository access only. Signing in at that address signs in the owner, so management permissions on this user would do nothing.

Why: the user is the build user that shares the owner’s sign-in address. Signing in at that address signs in the owner, so a management permission on it would do nothing. Nothing was changed.

Fix: give the management permission to a member who has their own sign-in address. Turning a permission off on this user is not refused.

Status 409, from the CloudRepo API, when you invite an address that already has an invitation waiting.

What you see:

This address already has a pending invitation with different access. Revoke it, then invite again.

The response also carries the waiting invitation’s ID in invitation_id.

Why: the address has a pending invitation, and the new one asks for different access. Nothing was stored or sent.

Fix: revoke the pending invitation, then invite again. On the Users page, Revoke and invite again does both.

Status 409, from the CloudRepo API, when the owner asks for the password builds list by email.

Why: the CloudRepo environment you called does not send this email to the owner’s address for your organization. A staging environment, for example, sends it only for CloudRepo’s own test organizations. Nothing was sent.

Fix: ask for the email from your organization’s portal in production.

Status 409, from the CloudRepo API, when the owner asks for the password builds list by email.

Why: no build has signed in with a password in the last 7 days, so there is nothing to list and no email was sent. The request did not use up the hour that limits how often the email goes out.

Fix: nothing to do. Ask again after a build that signs in with a password has run.

Status 409, from the CloudRepo API, when the owner turns password builds off. The code is in the code field, and status reads confirmation_required.

What you see: one of these sentences, in message:

Builds for 3 members still sign in with a password.
No build has signed in with a password in the last 30 days.

The first names the number of members. The response also lists the first page of those members and the counts of what turning password builds off would stop.

Why: this is the first step. The request carried no Confirmation header, so CloudRepo changed nothing and told you what the switch would stop.

Fix: read what it says. To go ahead, send the request again with the owner’s confirmation in the Confirmation header. The portal does both steps for you: it shows this list, then asks the owner to confirm.

Status 409, from the CloudRepo API, when you create or rotate a service key for a repository.

What you see:

This repository is being deleted. Nothing else can be changed until that finishes.

Why: the repository the key is for is being deleted, so CloudRepo makes no key for it. Nothing was created.

Fix: nothing to do for this repository: the deletion has to finish first. Make the key for another repository.

Status 409, from the CloudRepo API, when you ask to turn on cleanup for a Docker repository.

What you see:

Cleanup isn't available for this repository yet.

Why: turning on cleanup is not open to your organization yet.

Fix: nothing to do. The request changed nothing, and every image and tag in the repository is as it was.

Status 410, from your repository host, when you run npm token create. The response’s code field carries the same code in upper case, NPM_CLI_TOKEN_CREATE_DISABLED.

Why: CloudRepo no longer creates tokens through the npm CLI. A token that npm token create made before reaches no repository, so CloudRepo refuses it at every install and publish.

Fix: in the portal, open Repository Tokens and create a repository token for the repositories npm uses. Put it in your .npmrc in place of any token npm token create made.

Status 410, from your repository host, when you run npm login --auth-type=legacy. The response’s code field carries the same code in upper case, LEGACY_NPM_LOGIN_DISABLED.

Why: npm’s legacy sign-in sends an email address as the user name, which current npm versions do not allow.

Fix: run npm login without --auth-type, which signs in through your browser. For a build, put a repository token in your .npmrc instead.

Status 410, from the CloudRepo API, when you set another member’s password.

What you see:

You can't set another member's password. Reset it from their page instead, and they'll get an email to choose a new one.

Why: nobody sets a password for someone else any more. This request used to, and it is retired.

Fix: use Reset password under Sign-in on the member’s page. CloudRepo emails the member a link to choose their own password.

Status 413, from the CloudRepo API.

What you see:

The request body is larger than this endpoint accepts.

Or, for a body with too many items, a message that names the most JSON tokens the endpoint accepts.

Why: the request body is bigger than this endpoint accepts.

Fix: send less in one request. For an action on many items, split it into several requests.

Status 413, from your repository host, when Docker pulls an image through a Docker proxy repository. The code is in the response’s code field, beside limit_bytes, the most CloudRepo accepts, and observed_bytes, the size it received.

Why: the upstream registry sent an image manifest larger than 4 MiB (4,194,304 bytes), the most a Docker proxy repository accepts. CloudRepo did not cache or serve it.

Fix: pull this image from its upstream registry directly. A Docker proxy repository cannot serve a manifest this large.

Status 415, from the CloudRepo API.

What you see:

Send the request body as JSON, with Content-Type: application/json.

Why: the request had a body, but did not say it was JSON.

Fix: send the body as JSON with the header Content-Type: application/json. With curl, add -H 'Content-Type: application/json'.

Status 429, from the CloudRepo API, when the owner asks for the password builds list by email.

Why: the email already went out for your organization in this hour of the clock, in UTC. CloudRepo sends it once an hour for each organization.

Fix: ask again after the hour changes. The email already sent has the list.

Status 429, from the CloudRepo API, when you delete a folder that is a Maven version, or restore a Maven version’s .pom file from the Trash.

What you see: in the portal’s Trash, “CloudRepo is busy updating this repository’s Maven metadata. Try the restore again in a moment.”

Why: after a version is deleted or restored, CloudRepo updates its artifact’s maven-metadata.xml, and too many of these updates are waiting right now. Nothing was deleted, and nothing was restored.

Fix: wait the seconds the Retry-After header names (20), then send the same request again. It is safe to retry. In the portal, wait a moment and press Restore again.

Status 500, from the CloudRepo API, when an owner or an Admin resets a member’s password.

What you see: the code password_not_ended, with no sentence.

Why: CloudRepo could not end the member’s current password, so it did not send the email. The password is as it was. The member’s sessions may have ended.

Fix: press Reset password again. If it fails again, contact support.

A proxy repository or a group asked an upstream registry for a file, and the upstream’s answer could not be used. CloudRepo does not record the file as missing, so the next request asks again. These responses carry the code in the code field, beside a sentence that says what happened. None of them carries Retry-After. One is about a path CloudRepo never asks an upstream for: group_custom_upstream_not_asked. One 502 is about an email, not an upstream, and comes last in this section: email_not_sent.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: the upstream’s host name points to a private or internal address, and CloudRepo only connects to public hosts.

Fix: set the proxy’s upstream URL to a host on the public internet.

Status 502, from your repository host, when npm installs through a proxy.

What you see:

The upstream named a download address CloudRepo will not use, so CloudRepo did not fetch this version.

Why: the upstream’s package document points this version’s download at an address CloudRepo will not fetch from.

Fix: pick another version of the package, or ask the upstream’s owner to fix the download address.

Status 502, from your repository host, when npm reads from your own upstream URL.

Why: the upstream answered in a way CloudRepo cannot use: a redirect CloudRepo does not follow, an error status, or a page that is not an npm package document. CloudRepo never reports this as “not found”, so npm does not look for the name somewhere else.

Fix: check the proxy’s upstream URL and login, and that the upstream serves npm packages at that address.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

What you see:

The upstream answered with a web page, not the file, such as a sign-in or maintenance page. Check this proxy's upstream URL and login.

Why: the upstream sent back a web page instead of the file.

Fix: check the proxy’s upstream URL and login. Open the URL in a browser to see the page it shows.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: the upstream refused the login saved on the proxy.

Fix: update the upstream login in the proxy’s settings, and check that the account can read the upstream.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: the upstream asked for a login, and the proxy has none saved.

Fix: set the upstream login in the proxy’s settings.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: the upstream redirected CloudRepo to an address it does not follow, such as another port, an internal address, or a long chain of redirects.

Fix: set the proxy’s upstream URL to the address the files are really served from.

Status 502, from your repository host, when a proxy reads from its upstream.

Why: the upstream’s TLS certificate is not trusted, so CloudRepo stopped before sending anything to it.

Fix: give the upstream a certificate from a public certificate authority, or use an upstream URL that has one.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: the upstream answered with an HTTP status other than the file or “not found”. The response names the status.

Fix: check the proxy’s upstream URL, and look up that status in the upstream’s own documentation.

Status 502, from your repository host, when a proxy reads from your own upstream URL.

Why: CloudRepo cannot use the upstream URL saved on the proxy, so it did not ask the upstream.

Fix: save the proxy’s upstream URL again, as a plain https:// address.

Status 502, from your repository host, when you read through a group.

Why: a proxy in the group could not ask its upstream.

Fix: try again in a few seconds.

Status 502, from your repository host, when Maven reads through a group that holds a proxy for your own registry. The code is in the code field.

What you see:

CloudRepo never asks a custom upstream for a path under api/ or service/, so this group doesn't ask the repositories after it for that path.

Why: the package’s groupId starts with api or service as a whole first part, such as service.acme. Registries serve their own API under those paths, so CloudRepo never asks your registry for them. The group then stops at that proxy and asks none of the repositories after it, so a build never gets another repository’s package in place of yours.

Fix: publish the package to a hosted repository in the group, or move the repository that holds it ahead of the proxy in the group.

Status 502, from the CloudRepo API, when an owner or an Admin resets a member’s password. The code is in the error field.

What you see:

The password-reset email didn't send. The member's old password has already stopped working. Try again.

Why: CloudRepo ended the member’s old password, then could not send the email with the link to choose a new one.

Fix: press Reset password again to send a new email. The member’s old password stays ended until they choose a new one.

These are temporary. Wait, then try again: when the response carries Retry-After, wait that many seconds. ddb_unavailable, repository_token_identity_unavailable and sso_config_unavailable carry Retry-After: 5, membership_lookup_unavailable carries no Retry-After, and sso_users_busy carries Retry-After: 10. The other entries below say their own.

Status 503, from your repository host or the CloudRepo API.

Why: CloudRepo could not read or write its data store for this request.

Fix: wait a few seconds and try again. If it keeps failing, contact support@cloudrepo.io.

Status 503, from your repository host, when your build sends a repository token.

Why: CloudRepo could not look up whose token it is, so it did not decide whether to accept it. Your token was not refused.

Fix: run the build again. If it keeps failing, contact support@cloudrepo.io.

Status 503, from the CloudRepo API, when you list a member’s tokens or change a member’s access.

Why: CloudRepo could not read your organization’s member list for this request.

Fix: try again in a moment. If it keeps failing, contact support@cloudrepo.io.

Status 503, from your repository host or the CloudRepo API.

What you see, from Maven, npm, pip or Docker:

SSO configuration is temporarily unavailable. Please retry shortly.

Some of these responses carry no sentence, and a build tool may show only the status, 503. From the CloudRepo API, for example when you create or rotate a repository token, add a user, accept an invitation, or sign in, the response carries the code and no sentence.

Why: CloudRepo could not read your organization’s sign-in settings for this request.

Fix: wait a few seconds, then try again. If it keeps failing for more than a few minutes, contact support@cloudrepo.io with the time it happened.

Status 503, from the CloudRepo API, when your organization’s member list is requested with GET /api/admin/sso/users. The portal asks for it for User Enrollment Status and for a member’s page. The response carries Retry-After: 10, the code in the error field and no sentence.

Why: The member list is busy. CloudRepo can read only a few pages of it at once, and none was free for 10 seconds, so yours did not start. Nothing is wrong with your organization or your account.

Fix: wait the number of seconds in the Retry-After header, then ask for the same page again. If it keeps failing for more than a few minutes, contact support@cloudrepo.io with the time it happened.

Status 503, from your repository host, when a proxy reads from its upstream. The code is in the code field, and the response carries Retry-After: 1.

Why: the proxy’s upstream repository could not answer just now.

Fix: try again in a moment. Your build tool may retry it on its own.

Status 503, from your repository host, when a proxy reads from your own upstream URL. The code is in the code field, and the response carries Retry-After: 60.

Why: CloudRepo could not use the upstream login saved on the proxy, so it did not ask the upstream.

Fix: set the upstream login again in the proxy’s settings.

Status 503, from your repository host, when a proxy reads from its upstream. The code is in the code field, and the response carries Retry-After: 1.

Why: CloudRepo was answering too many upstream reads at once, so it did not ask the upstream.

Fix: try again in a moment.

Status 503, from your repository host, when npm reads a package through a proxy. The code is in the code field, and the response carries Retry-After: 1.

Why: the registry was answering too many reads at once.

Fix: try again in a moment. npm retries this on its own.

Status 503, from your repository host, when pip or uv reads a Python index. The code is in the code field, and the response carries Retry-After: 1.

Why: the repository was answering too many index reads at once.

Fix: try again in a moment. pip and uv retry this on their own.

Status 503, from your repository host, when you upload a Python package. The code is in the code field, and the response carries Retry-After: 1.

Why: the repository was answering too many requests at once, so CloudRepo could not check the package name before it stored your upload.

Fix: upload again in a moment.

Status 503, from your repository host, when Maven or Gradle reads through a group. The code is in the code field, and the response carries Retry-After: 1.

Why: a repository in the group could not be read just now. CloudRepo does not answer “not found”, so your build does not fetch the file from somewhere else.

Fix: try again in a moment. Maven and Gradle retry this on their own.

Status 503, from the CloudRepo API, when you browse a repository’s files. The response carries Retry-After: 1.

What you see:

This folder could not be read to its end just now. Please try again in a moment.

Why: CloudRepo could not read the whole folder this time.

Fix: open the folder again in a moment.

Status 503, from the CloudRepo API, when you make a Docker proxy repository public.

What you see:

We could not check this proxy's cache just now. Nothing changed. Try again in a minute.

Why: before a Docker proxy becomes public, CloudRepo checks that its cache is empty, and it could not read the cache.

Fix: try again in a minute. Nothing was changed.

Status 503, from the CloudRepo API, when you make a Docker proxy repository public.

What you see:

We could not check this repository's upstream credentials just now. Nothing changed. Try again in a minute.

Why: before a Docker proxy becomes public, CloudRepo checks that no upstream login applies to it, and it could not read that.

Fix: try again in a minute. Nothing was changed.

Status 503, from the CloudRepo API, when you ask to become an Admin.

What you see:

We couldn't send your request. Try again in a minute.

Why: CloudRepo could not send your request to the organization’s owner.

Fix: try again in a minute.

Status 503, from the CloudRepo API, when you turn off Require MFA for the organization. The response carries Retry-After: 5.

Why: CloudRepo could not read what it checks before it turns the requirement off, so nothing changed.

Fix: try again in a few seconds.

Status 503, from the CloudRepo API, when you open the list of builds that still sign in with a password. The response carries Retry-After: 5.

Why: CloudRepo could not read the list just now.

Fix: try again in a few seconds.

Status 503, from the CloudRepo API, when the owner asks for the password builds list by email. The response carries Retry-After: 5.

Why: CloudRepo could not read the list or send the email.

Fix: try again in a few seconds.

Status 503, from the CloudRepo API, when you turn password builds off. The response carries Retry-After: 5.

Why: CloudRepo could not check which builds still use a password, so it changed nothing.

Fix: try again in a few seconds.

Status 503, from the CloudRepo API, when you turn password builds off or on. The response carries Retry-After: 5.

Why: CloudRepo could not save the change.

Fix: try again in a few seconds, then check the setting on the page.

Docker answers in the registry’s format: a JSON body with a list of errors, each with an upper-case code and a message. The Docker CLI prints the code in lower case, with spaces for underscores.

Status 401, from your repository host.

What you see, from docker login:

unauthorized: Docker takes a repository token or a service key, not a member password, and refuses a token that is revoked, expired or from another organization: https://www.cloudrepo.io/docs/authenticate/repository-tokens.html#use-the-token-in-your-client

Why: Docker sent no credential, or one CloudRepo did not accept: a password, a revoked or expired token, or a mistyped one. The answer is the same for each, so it names none of them.

Fix: sign in to your organization’s host with your email address and a repository token. CloudRepo’s registry does not check the user name when the password is a repository token. Pipe the token in:

Terminal window
echo "$CLOUDREPO_TOKEN" | docker login your-org.mycloudrepo.io --username you@example.com --password-stdin

Status 403, from your repository host.

What you see Why Fix
denied: insufficient permissions Your credential does not reach this repository, or you pushed or deleted with one that cannot write to it. Use a repository token that reaches this repository, with write scope to push or delete.
denied: requested scope not authorized for this token The repository token you signed in with does not reach this repository, or the repository does not exist. With an access key, the subscriber or group that owns it has not been given this repository. Check the repository’s name, then use a repository token that includes it. With an access key, ask the organization that gave you the key for access to this repository.
denied: the upstream registry of group member <member> asked for a login or refused access, so it may not hold this image. The group stops here and serves nothing from the members after it. Check the image name, move the member that holds the image above <member>, or give <member> an upstream credential that can read it. A group asks its members in order. It stops at the first member that answers with an error, and serves nothing from the members after it. This member’s registry asked for a login or refused access. Some registries, such as Quay, answer this way for an image they do not hold. Check the image name. If another member holds the image, move that member above this one in the group. If this member’s registry should hold it, give the member an upstream credential that can read it. A public group serves only its public members, to everyone, signed in or not.

Status 400, from your repository host.

What you see Why Fix
name invalid: Docker image names require at least two segments: <repository>/<image> The image reference has no repository in it, or no image after the repository. Name the image under its repository: your-org.mycloudrepo.io/repositories/your-repo/my-app:1.0.0.
name invalid: invalid image name A part of the image name is not lower-case letters and digits, joined by ., _ or -. Docker image names can be up to 255 characters. Rename the image, for example my-app rather than My_App/.
name invalid: invalid repository name The repository part of the reference is not a valid repository name. Check the repository name in the portal and use it exactly.
name invalid: Name the image <your-host>/repositories/<repository>/<image>:<tag>. Put repositories/ and your repository name after the host. The reference leaves out repositories/ and the repository after the host, as your-org.mycloudrepo.io/your-repo/my-app and your-org.mycloudrepo.io/my-app do, or it names no image after the repository. Where this page shows <your-host>, the message carries your own host. Name the image with repositories/ and your repository name after the host: your-org.mycloudrepo.io/repositories/your-repo/my-app:1.0.0.
name invalid: invalid tag format The tag does not start with a letter, a digit or _, holds a character other than letters, digits, ., _ and -, or is longer than 128 characters. Use a tag such as 1.0.0 or latest.

Status 404, from your repository host.

What you see Why Fix
name unknown: repository not found No Docker repository in your organization has that name. Check the repository name in the reference, after /repositories/.
name unknown: group repository has no members The group has no member repository it can serve to you: it has none, or you cannot read any of them. Add a member repository to the group in the portal, or pull with a credential that can read its members.

Status 404, from your repository host.

What you see, from the Docker CLI:

manifest unknown: manifest not found

Why: the repository has no image with that tag or digest. For a group repository, none of its members has it.

Fix: check the tag and the repository name in the image reference.

Status 404, from your repository host.

Why: Docker asked for a layer the repository does not hold. During a push this is normal: Docker checks for each layer before it uploads it. During a pull, the image’s manifest names a layer that is missing.

Fix: during a pull, push the image again so every layer is uploaded.

Status 400, from your repository host, when you push an image.

Why: the image’s manifest names a layer that was never uploaded to this repository.

Fix: push the image again.

Status 400, from your repository host, when you push an image.

What you see Why Fix
manifest invalid: unsupported manifest media type The client sent a manifest type CloudRepo does not store. Push with the Docker CLI or another OCI client that sends a Docker or OCI image manifest.
manifest invalid: manifest exceeds maximum size The manifest is larger than CloudRepo accepts. Push an image with fewer layers or platforms in one manifest.
manifest invalid: manifest body is not valid JSON or computed digest does not match reference The manifest arrived damaged, or does not match the digest it was pushed under. Push again.

Status 400, from your repository host.

Why: a layer’s content does not match its digest, so it arrived damaged, or a digest in the request is not a sha256: digest.

Fix: push again. Use sha256 digests.

Status 400, from your repository host, when you push an image.

Why: one layer of the image is larger than CloudRepo accepts.

Fix: make the layer smaller, for example by splitting a large build step into several. If you cannot, contact support@cloudrepo.io.

Status 404, from your repository host, when you push an image.

Why: Docker continued a layer upload that CloudRepo no longer has, for example after a long pause or a failed upload.

Fix: push again. Docker starts a new upload.

Status 400, from your repository host, when you push an image.

Why: Docker sent a piece of a layer with no content.

Fix: push again.

Status 405, or 400, from your repository host.

What you see Why Fix
unsupported: Writes are not permitted to group repositories You pushed to a group repository. A group has no storage of its own. Push to one of the group’s hosted member repositories. The image appears through the group at once.
unsupported: method not allowed The client used a request method this part of the registry does not take, such as deleting a single layer. Use the requests the registry supports: pull, push, and deleting an image’s manifest.
unsupported: catalog not yet implemented The client asked for the registry’s catalog of repositories, which CloudRepo does not serve. List your repositories in the portal.

Status 500, or 502 from a group repository, from your repository host.

What you see Why Fix
unknown: internal error CloudRepo failed unexpectedly while answering Docker. Try again. If it keeps failing, contact support@cloudrepo.io with the time of the failure.
unknown: the upstream registry of group member <member> answered an error A group asks its members in order. It stops at the first member that answers with an error, and serves nothing from the members after it. This member’s registry answered with an error, such as a server error or a page that is not an image. Try again in a few minutes. If it keeps failing, contact support@cloudrepo.io with the time of the failure and the member’s name.

Status 429, from your repository host. Each answer carries Retry-After.

What you see Why Fix
toomanyrequests: Too many failed sign-ins from your network. Try again in 60 seconds. Your network sent more than 5 different wrong passwords for one member to docker login within a minute, so CloudRepo refuses further passwords for that member from your network. Retry-After is 60. Wait 60 seconds, then sign in again with the right credential. Where you can, use a repository token in place of a password.
toomanyrequests: this server is answering too many manifest reads at once, retry shortly CloudRepo was already reading as many manifests as it allows at once. Retry-After is 1. Pull again after a second.
toomanyrequests: this server is answering too many tags lists at once, retry shortly CloudRepo was already answering as many tags lists as it allows at once. Retry-After is 1. List the tags again after a second.

Status 503, from your repository host.

What you see Why Fix
unavailable: the registry is busy; please retry You pushed an image manifest while CloudRepo was already processing as many manifests as it allows at once. Retry-After is 1. Push again after a second.
unavailable: This repository changed while the layer was downloading. Retry the pull. You pulled a layer through a Docker proxy that signs in to its upstream, and the proxy changed while the layer downloaded (it became public, or it began to be deleted), or CloudRepo could not confirm it had not. CloudRepo discarded the layer. Pull again.
View as Markdown