Repository navigation
v0.1.48
v0.1.48 adds required MFA, recovery codes and passkeys, a password policy and session lifetimes, station data plans for Wi-Fi or wired and cellular stations, SIM providers and a SIM inventory, a Data Retention settings tab, a customizable dashboard with four new charts, avatars for drivers and CSMS users, a choice of the sites and stations AI insights analyze, audit entry details and real log uploads in the simulator. It also carries a large security hardening pass, with breaking changes to permissions, sign-in proofs, cards, OCPI roles and deployment defaults. Read the breaking changes and upgrade notes before you upgrade from v0.1.47.
Breaking changes
Permissions
- Payment provider settings need the Payment settings permission.
GETandPUT /v1/settings/stripe,/v1/settings/adyenand/v1/settings/payments, with their test and webhook routes, now needsettings.payment:read(reads) orsettings.payment:write(everything else) instead ofpayments:readorpayments:write. Affected: custom roles and API keys that managed payment providers with Payments permissions. Admins keep access: the upgrade grants both permissions to every user with the admin role. Action: grant Payment settings to any other user or API key that manages payment providers. See Permissions. - Station commands need the permission of their feature. The generic OCPP command routes (
POST /v1/ocpp/commands/{v16,v21}/<action>), the station configuration push, the station charging profile push and clear, and the station certificate install and delete now also need the write permission of the matching feature: firmware commandssettings.firmware:write; configuration, network, monitoring,DataTransferand any command not listedsettings.stationConfig:write; certificate commands and aSign*Certificatetriggercertificates:write(a Plug and ChargeDataTransferneeds both); charging profile and grid signal commands (UpdateDynamicSchedule, the DER controls, AFRRSignal, UsePriorityCharging), the charging profile refresh and the composite schedulesmartCharging:write; reservation commandsreservations:write; start, stop, RequestBatterySwap and the station stop of an active sessionsessions:write; tariff and cost commands (SetDefaultTariff, ClearTariffs, ChangeTransactionTariff, CostUpdated)pricing:write; CustomerInformationdrivers:write. A security profile change throughPATCH /v1/stations/:idor the simulator station PATCH needssettings.stationConfig:write. Reset, availability, unlock, cache, local list, display message, the Get reads and triggers that sign nothing still need onlystations:write, and so do station password changes and credential rotation. Affected: the default operator role, which has neithersettings.firmwarenorsettings.stationConfig: it loses UpdateFirmware, SetVariables, ChangeConfiguration and SetNetworkProfile from the station OCPP Commands tab and the station Configurations tab, and the monitoring rules on the Charging Needs tab. Also custom roles and API keys with onlystations:write. Admins keep access through the upgrade grant. Action: grant the feature permission to each operator who sends those commands. The default operator role does not gainsettings.stationConfig:write. See Permissions. - Station log uploads need Station configuration write.
GetLog(OCPP 2.1 and 1.6) andGetDiagnostics(OCPP 1.6) needsettings.stationConfig:writeon top ofstations:write, because they upload the station logs to a URL the caller picks. Affected: operators and API keys that fetched station logs withstations:writealone. Action: grantsettings.stationConfig:writeto the people who fetch station logs. - Support case refunds need Payments write.
POST /v1/support-cases/:id/refundnow needspayments:writein addition tosupport:write, and checks the site of the session. Affected: custom roles with Support write and without Payments write. Action: grant Payments write to staff who refund from a case. The CSMS hides the refund button without it. - Free vend needs Payments write.
POST /v1/sites/:id/free-vendneedspayments:writeon top ofsites:write, a site-restricted user cannot turn it on at a site with a platform fee, and the generic sitePATCHrefusesfreeVendEnabled. Affected: custom roles with Sites write and without Payments write, and site-restricted operators. Action: grant Payments write, or have a user with access to every site turn it on at a site with a fee. See Free Vend. - Narrower read permissions for audit history, money and reports.
GET /v1/auditand/v1/audit/:entityType/:entityIdreturn setting rows only withsettings.system:read; user, role and API key rows only withusers:read; payment, invoice and fleet rows only withpayments:read; driver, token and vehicle rows only withdrivers:read; report schedule rows only withreports:read; AI conversation rows only to their owner. Fleet billing mode, credit limit and member opt-out writes needpayments:writetoo, and the credit limit readpayments:read. Site and station revenue figures are null withoutpayments:read. Revenue and NEVI reports needpayments:read, the driver activity reportdrivers:read, on generate, on schedule and at each scheduled run. The sessions report leaves out the driver columns withoutdrivers:readand the money columns withoutpayments:read, and the driver activity report leaves out spend withoutpayments:read, so column positions shift in those files. Affected: the default viewer role (audit rows), custom roles, and scripts that parse report files by column position. Action: grant the read permission, and read report columns by header name. See Reports and Audit Logs. - Stored reports follow the reader's permissions. Each stored report records whether it holds driver columns and money columns, from the permissions of the request that generated it (an API key's scope included) or, for a scheduled run, of the schedule owner. The report list hides it, and detail and download answer 404, for a reader without
drivers:readorpayments:readwho may not see those columns. Affected: users who downloaded reports another user generated with those columns. Action: generate the report yourself. It leaves out the columns you may not see. - Security notification templates need System settings write. Saving or deleting the password reset, invitation, email verification, sign-in code, payout onboarding and every security notice template through
PUTorDELETE /v1/notification-templatesneedssettings.system:write(403INSUFFICIENT_PERMISSIONS). A save must keep the shipped link and code variables, may hold a link only as a wholehrefor as text, and may link only to the CSMS or portal address (400VALIDATION_ERROR). A stored security template that breaks these rules is not sent: the shipped template goes out instead.PUT /v1/ocpp-event-settingsrefuses an event type that is not an OCPP event. Affected: the default operator role, which hasnotifications:writeonly, and operators with customized security templates. Action: edit security templates as a user with System settings write, and keep their links and codes. The CSMS shows them read-only without it. See Editing a Template. - Stored secrets need System settings read, and never follow a new address. The Plug and Charge and reCAPTCHA settings reads return their secrets as null plus a
*Configuredflag unless the caller holdssettings.system:read.PUT /v1/pnc/settingswith a changed Hubject base or token URL and no new client secret clears the stored secret.PATCH /v1/ocpi/partners/:idwith a newversionUrland no token clears the registration token. Affected: scripts that change those URLs without sending the secret again, and scripts that read the secrets with another permission. Action: send the secret with a URL change, and read secrets with System settings read.
Sign-in, accounts and MFA
- Switched-off MFA methods apply at sign-in and renewal. Turning a method off in Settings > Security > MFA used to stop only new setups. Now a code of a switched-off method is refused with 403
MFA_METHOD_DISABLED. Accounts with another factor (a passkey or recovery codes) sign in with it. Accounts whose only factor is switched off prove it once more, then set up an allowed method. Action: before you upgrade, check which methods are off and whether users still rely on them. See Switched-Off Methods. - Adding a factor needs a fresh proof.
POST /v1/users/me/mfa/confirm,POST /v1/portal/driver/mfa/confirmand passkey registration called with a session needproof: a factor the account holds, else the current password, or a recent SSO sign-in for an SSO account without either. Without it: 400VALIDATION_ERRORwith the accepted kinds inproofKinds. Action: scripts and custom clients that set up MFA send the proof. See Confirming Security Changes. - Turning MFA off and removing a passkey need a factor or a recovery code.
DELETE /v1/users/me/mfaandDELETE /v1/portal/driver/mfatake{ proof }(a factor the account holds or a recovery code) instead of{ password }. Removing a passkey while the account holds a factor no longer accepts the password. A request with the password alone gets 400VALIDATION_ERRORwithproofKinds. Action: update any script that turns MFA off. The driver app sends drivers to the portal for this change. - API keys created from a session need a fresh proof.
POST /v1/api-keys, and aPATCH /v1/api-keys/:idthat widens a key's permissions or sites, take the sameproofwhen called with a session. Calls made with an API key are unchanged. Every new key emails its owner. See API Keys. - User administration and SSO settings need a fresh proof from a session. A signed-in session sends
proof(the current factor, else the password) to:PUT /v1/sso/settings;POST /v1/userswith an admin-level permission set;PATCH /v1/users/:idwhen it changes the email or phone of an admin-level user, reactivates an admin-level user, changes a role so that the user gains permissions, or turnsssoLinkAllowedon;PUT /v1/users/:id/permissionswhen it adds a permission; andPOST /v1/users/:id/reset-password,/resend-inviteand/mfa/resetfor another user. Without it: 400VALIDATION_ERRORwithproofKinds. Admin-level means every permission, or any ofusers:write,settings.security:write,settings.apiKeys:write,settings.system:writeandsettings.payment:write.POST /v1/users/:id/reset-passwordand/resend-inviteon your own account answer 403SELF_EDIT_FORBIDDEN(use/v1/users/me/change-password).PUT /v1/sso/settingsrefuses adefaultRoleIdwhose role grants more than the editor holds (403PERMISSIONS_EXCEED_OWN) and an unknown role (400). Requests made with an API key need no proof, but an API key cannot change the identity provider and account mapping fields ofPUT /v1/sso/settings(entryPoint,issuer,cert,attributeMapping, or anallowedDomainschange that clears or widens the list): 403SESSION_REQUIRED. Affected: scripts that call these routes with a session cookie or session token, and API keys that change the SSO identity provider. Action: sendproofor use an API key, and change the identity provider from a signed-in session. See Confirming Security Changes. - Own email and phone are not editable through the users route.
PATCH /v1/users/:idon your own account withemailorphoneanswers 403SELF_EDIT_FORBIDDEN. Change your own phone throughPATCH /v1/users/me, which asks for a proof when the account holds a factor, and a code sent to the new number when the account signs in with SMS codes. An admin still changes other users' email and phone on the Users page. See Your Own Account. - Phone numbers are validated and stored in international format. Every route that stores a phone (
PATCH /v1/users/me,POST /v1/users,PATCH /v1/users/:id,POSTandPATCH /v1/drivers/:id,PATCH /v1/portal/driver/profile,POST /v1/portal/auth/register) answers 400VALIDATION_ERRORfor a number that does not parse. A number without a country code is read as North American. Existing rows are not rewritten. Action: integrations that send local formats from other countries send the country code. See Phone Number Changes. - SSO signs in to an existing password account only after a link. SSO signs in an existing account only when its identity provider subject matches the one stored at its first SSO sign-in. A password account that never signed in by SSO is refused (
/login?error=sso_link_required) until an admin turns on "Allow SSO sign-in" on the user, or the user signs in with the password once and confirms the link. Accounts that SSO created before the upgrade are linked at their next SSO sign-in. After an admin clears an account's SSO link, the next SSO sign-in needs the link allowed again. Affected: operators who sign in by SSO to an account created with a password. Action: allow or confirm the link. See SSO Accounts. - The CSMS and the portal cannot be shown in a frame. Both images send
Content-Security-Policy: frame-ancestors 'none'andX-Frame-Options: DENY. Affected: sites and kiosks that embed the portal or the CSMS in an iframe. Action: open them in their own window or tab.
Drivers, cards and payments
- Vehicle identifiers cannot be registered as cards.
POST /v1/portal/tokensrefuses the typesVIN,EVCCIDandMacAddressand any value shaped like a vehicle id (a MAC address in any separator form, a 17-character VIN, aVID:prefix) with 400VALIDATION_ERROR, and a value that is another driver's vehicle id with 409VEHICLE_ID_IN_USE. Operator card create, update and import refuse a value that is another driver's vehicle VIN or EVCCID (409VEHICLE_ID_IN_USE, an import reports the row), and setting a vehicle id that another driver's card holds answers 409VEHICLE_ID_IN_USE. Values compare without theVID:prefix, separators or case. A 1.6VID:idTag matches only cards of a vehicle id type, and any other card matches its exact spelling, as in v0.1.47. Turning Autocharge on for a vehicle needs the vehicle id to have been seen in one of the driver's own sessions, else 409AUTOCHARGE_CLAIM_REQUIRED. Operators can still turn it on for a driver. Affected: drivers on the portal and the mobile app, and operators and API clients that create or import cards. Action: drivers use Use This Vehicle for Plug-In Charging on a session first. See Enrolling a Vehicle and Autocharge in the portal. - Prepaid balance changes need the balance the client read.
PATCH /v1/tokens/:idwithprepaidBalanceCentsanswers 400VALIDATION_ERRORwithoutexpectedPrepaidBalanceCents(the balance the client read, null for a postpaid card), and 409PREPAID_BALANCE_CONFLICTwhen that value is no longer current.POST /v1/tokensthat re-adds a removed card of the driver sets the sent balance only while the card holds none (409PREPAID_BALANCE_CONFLICT). Affected: API scripts and integrations that set a prepaid balance. Action: read the card and send itsprepaidBalanceCentsasexpectedPrepaidBalanceCents. The CSMS already sends it. See Prepaid Cards. - Manual fleet and member invoices for a month that has not ended are refused. Fleet and member invoice generation for the current month answers 400
VALIDATION_ERROR, and the CSMS Generate Invoice picker lists ended months only. Fleet credit limit notices and invoice run digests go to users with access to every site only. Affected: operators and API clients that invoiced the running month. Action: invoice a month after it ends. See Fleets. - Push tokens are deleted whenever a driver's sessions end. Every password change or reset, MFA or passkey change, sign-out everywhere and deactivation deletes all of the driver's push tokens, the token of the device that made the change included. Driver app builds older than v0.1.48 do not register their token again after such a change, so that device gets no push notifications until the driver signs in again. Affected: drivers on older app builds. Action: ship the v0.1.48 driver app, which registers its token again after these changes.
Stations and OCPP
- Configuration templates cannot hold the values the data plan sends at every boot. Creating, editing or pushing a template that names the heartbeat interval or a meter value setting (
HeartbeatInterval,MeterValueSampleInterval,MeterValuesSampledData,StopTxnSampledData,ClockAlignedDataIntervalandMeterValuesAlignedDataon 1.6;OCPPCommCtrlr.HeartbeatInterval,SampledDataCtrlr.TxUpdatedInterval,TxUpdatedMeasurands,TxEndedMeasurands,AlignedDataCtrlr.IntervalandMeasurandson 2.0.1 and 2.1) answers 400DATA_PLAN_SETTING_IN_TEMPLATE, and the editor names the refused rows. Affected: saved templates that hold one of these values, such as the earlier seeded demo templates withHeartbeatInterval. They cannot be pushed until the rows are removed. Action: remove those rows, and set the values in Settings > Integrations & Features > OCPP and the station data plan. A single command from the station's OCPP Commands tab can still set them until the next boot. See Templates. - Station image uploads refuse SVG and use a form upload.
POST /v1/stations/:id/images/upload-urlanswers 400 forimage/svg+xml, and returns a size-limited upload form (uploadUrlplusfields) instead of a PUT URL. The client posts the fields and the file, then confirms with the returneds3Key, and the confirm checks and re-encodes the image. Existing SVG images download as attachments. Affected: operators who uploaded SVG images, and API clients that PUT to the upload URL. Action: upload PNG, JPEG, GIF or WebP, and post the form fields. See Upload. - Security profile 3 needs a client CA, and Hubject no longer signs station TLS certificates. Without
OCPP_TLS_CAthe OCPP server refuses profile 3 connections and logs an error at start. With Hubject as the certificate provider, aChargingStationCertificatesigning request (the profile 3 client certificate) is no longer sent to Hubject: it stays pending for the operator (MANUAL_SIGNING_REQUIRED), and a returned certificate without the request's key is refused. Affected: installs that run profile 3 against Node's public root certificates, and Hubject installs that renewed station TLS certificates through SignCertificate. Action: setOCPP_TLS_CAto the CA that issues station certificates, and sign station certificates with the operator CA. See Security Profiles.
OCPI roaming
- New OCPI partners need role approval. A partner registered after the upgrade, inbound or outbound, reaches no OCPI module until an operator approves its roles on the partner page (Roaming > partner > Approved Roles). Every module now needs an approved role that serves it: a partner gets 403 on a module its approved roles do not serve (for example an NAP or NSP on sessions, or an OTHER, SCSP or PTP role on any module), and a remote start needs an approved EMSP or HUB role. Partners registered before the upgrade keep the roles they registered with (the OCPI role names of their own party), so a partner that used only modules of its declared roles keeps working, and one that used a module outside its roles gets 403. Affected: operators who connect new partners, and partners that used modules outside their declared roles. Action: approve the partner's roles after the handshake. See Partner Registration.
- OCPI registration tokens expire after 30 days. A registration token that a pending partner has not used within 30 days answers 401. Affected: handshakes in progress. Action: issue a new registration token.
Data retention
- Data retention moved to its own settings tab and routes. The generic settings routes no longer list or write the retention keys (400
VALIDATION_ERROR). UseGETandPUT /v1/settings/data-retention, or Settings > Data Retention in the CSMS. The audit history, sign-in history and station security event retention take 0 (keep forever) or 30 to 3650 days: the route and the seed--apply-configrefuse 1 to 29, and shortening one of them from a signed-in session needsproof(API keys need none). The Helm chart no longer writes retention and fails the render on anappSettingsretention value that differs from the earlier chart default. The CDK config refuses retention keys inappSettings. Affected: scripts that set retention through the generic settings routes, and Helm and CDK configs that set retention. Action: use the new route or tab, remove the retention values from your Helm values (with--reuse-values, pass--set appSettings.<key>=null) and from the CDKappSettings. See Where to Change Them.
System Information
- System Information reports the OCPP settings per OCPP process.
GET /v1/system/infono longer returnsnetwork.ocppPort,network.ocppHost,network.ocppHealthPort,network.ocppTlsPort,network.ocppTlsEnabled,rateLimits.ocppMaxConnectionsPerIp,rateLimits.ocppMaxMessagesPerIpPerSecondandocpp.instanceId. The API read those values from its own environment, which on Helm and AWS is not the OCPP environment. Each OCPP process now reports its own values inocppProcesses[].config(version,wsCompression,tlsEnabled,tlsPort,clientCaConfigured,maxConnectionsPerIp,maxMessagesPerIpPerSecond,maxPendingAuthPerIp,trustedProxiesConfigured), with itsinstanceId, connected stations and start time.configis null for an OCPP process of an earlier release. The reply also addsnetwork.ocppStationTlsUrl,security,simProviders,dataPlansand twosecretsflags. Affected: scripts and monitoring that read the removed fields. Action: readocppProcesses[].configinstead. See System Information.
AI insights
- The site update no longer turns AI insights on or off.
PATCH /v1/sites/:idwithaiInsightsEnabledanswers 400VALIDATION_ERRORnaming the new scope route. Which sites and stations automatic insights analyze is now one company-wide setting, the AI insights scope, written only throughPUT /v1/settings/ai/insights-scope(settings.system:writeand access to every site). Site replies still returnaiInsightsEnabled, the stored selection. Affected: scripts that switched insights per site with the site update. Action: sendPUT /v1/settings/ai/insights-scopewith{ "mode": "selected", "sites": [{ "siteId": "...", "selected": false }] }instead. See Automatic Insights settings.
Deployment
- Published images run in production mode. The API, OCPP, OCPI and worker images set
NODE_ENV=production, and an unsetNODE_ENVno longer allows the simulated payment provider. The API also defaultsNODE_ENVtoproductionwhen run from source without it. A test install that selected the simulated provider withoutPAYMENTS_ALLOW_SIMULATED=truenow answers 400PAYMENT_PROVIDER_NOT_CONFIGUREDon every paid start instead of charging nothing. Affected: deployments outside Helm and CDK (plaindocker run, your own Compose files or manifests) that relied on the simulated provider,CORS_ORIGIN=*or private AI base URLs. Action: setNODE_ENV=developmentandPAYMENTS_ALLOW_SIMULATED=truefor test installs, or configure a real payment provider. See Environment Variables. - Docker Compose needs generated secrets, and the Redis default user is off.
docker-compose.ymlhas no defaults forJWT_SECRETandSETTINGS_ENCRYPTION_KEY, sodocker compose upwithout them in.envstops with an error.scripts/docker-build.shwrites random values into.env, except for a stack whose database volume already exists: there it keeps the earlier built-in settings key and prints a warning, because a new key cannot decrypt the stored credentials. The Redisdefaultuser is off: host tools (npm run dev:*,redis-cli, integration tests) sign in as the new development-onlyadminuser withREDIS_ADMIN_PASSWORD. The OCPP Node inspector listens on127.0.0.1inside the container. Affected: every Compose install, including local stacks built before this release. Action: see the Compose steps under Upgrade notes. The seeded admin and driver accounts are unchanged. See Secrets and Redis Users. - The API refuses a known or short JWT secret and a trust-all proxy. The API refuses to start with a
JWT_SECRETshorter than 32 characters or equal to a published example value (outsideNODE_ENV=test). The API and OCPI refuse aTRUST_PROXY, and the OCPP server anOCPP_TRUSTED_PROXY_CIDRS, that contains0.0.0.0/0,::/0or an IPv6 range that holds every IPv4-mapped address (such as::ffff:0.0.0.0/96). Affected: deployments with the example secret, a short secret or a trust-all proxy entry. Action: generate a new secret (every user then signs in again) and list only your proxy addresses. See API Server. - Helm chart defaults and refusals.
css.enableddefaults tofalse. Pods receive only their own Secret keys. Every workload runs as non-root with a read-only root filesystem and anemptyDiron/tmp, and the migrate image runs as non-root. The chart refusesapi.env.nodeEnvother thanproduction;monitoring.enabledwithoutmonitoring.grafana.adminPasswordor withadmin;api.env.corsOrigincontaining*;ocpp.env.trustedProxyCidrsorapi.env.trustProxywith0.0.0.0/0,::/0or an IPv6 range that holds every IPv4-mapped address; and on install asecrets.jwtSecretorsecrets.settingsEncryptionKeyshorter than 32 characters or equal to a published example value. Affected: Helm users who run the simulator, write to the root filesystem from extra containers or mounts, or set those values. Action: setcss.enabled: truefor a test install, set a Grafana password, and fix the refused values. See Helm Chart. - Helm needs
api.env.trustProxybehind an HTTPS gateway. The chart refuses to render whilegatewayAPI.enabledis true,api.env.trustProxyis empty (the default), and the Gateway the chart creates has an HTTPS or TLS listener or the chart uses a Gateway it does not create. Without it, sign-in cookies had noSecureflag. Affected: Helm installs behind HTTPS that left the value empty. Action: setapi.env.trustProxyto the gateway pod range, for example10.244.0.0/16, before you upgrade. See Gateway API. - CDK refuses a weakened prod config and overrides of derived values. For
env: prod,waf.enabled: false,deletionProtection: false(database and load balancer) andecs.executeCommand: trueare refused unlessoverrides.<name>.reasongives the reason. In every environment, a serviceenvorsecretsentry for a value the stack sets itself (NODE_ENV,PAYMENTS_ALLOW_SIMULATED,SEED_DEMO,TRUST_PROXY,OCPP_TRUSTED_PROXY_CIDRS,CORS_ORIGIN,COOKIE_DOMAIN,JWT_SECRET,SETTINGS_ENCRYPTION_KEY) fails the config load. Affected: CDK users with a custom prod config or service overrides of those names. Action: remove the value or add the reason, and usepayments.allowSimulatedProviderandseedDemo.enabledinstead of the env overrides. See Security Posture. - Email templates refuse unescaped variables. HTML email bodies refuse
{{{variable}}}except{{{content}}}in the email layout (400VALIDATION_ERRORon save throughPUT /v1/notification-templates,PUT /v1/ocpp-event-settingswith the email channel andPUT /v1/settings/email.wrapperTemplate), and stored HTML bodies render such variables escaped. Affected: operators with customized email templates that use triple braces. Action: use{{variable}}. See Email Layout.
Upgrade notes
- Every user signs in once more. Cookie signing keys are now derived per use, so sessions signed with the earlier key are refused after the upgrade.
- Docker Compose installs:
- Run
scripts/docker-build.shonce. It writesJWT_SECRET,SETTINGS_ENCRYPTION_KEY,REDIS_ADMIN_PASSWORDand the five service Redis passwords (REDIS_API_PASSWORD,REDIS_OCPP_PASSWORD,REDIS_OCPI_PASSWORD,REDIS_WORKER_PASSWORD,REDIS_CSS_PASSWORD) into.envwhen one is missing or a published example value, and recreates Redis with the new passwords. - To edit
.envby hand instead, setJWT_SECRETand the Redis passwords to new random values (openssl rand -hex 32). Keep theSETTINGS_ENCRYPTION_KEYthe stack already uses (for a stack built before this release, the earlier built-in value): a new key cannot decrypt the stored SMTP, Twilio, payment and other credentials, and you would have to enter them again. - Point host tools at the
adminRedis user:redis-cli --user admin --pass "$REDIS_ADMIN_PASSWORD". - For a host debugger on port 9229, set
NODE_INSPECT=--inspect=0.0.0.0:9229in.env.
- Run
- Cards that match another driver's vehicle. The upgrade deactivates, once, every active driver card whose value equals another driver's vehicle VIN or EVCCID (the vehicle wins), marks it as turned off by the system for that reason, and marks the local authorization lists that hold it as changed. The migration logs the count. It writes no card history row and sends no driver notice and no OCPI push. Cards shaped like a vehicle id that match no other driver's vehicle stay active and match only their exact spelling, and the migration logs their count too. Review both counts. After checking, you can move such a card to a vehicle or reactivate it.
- OCPI partners. Partners registered before the upgrade keep their declared roles. Check each partner's Approved Roles card, and approve roles for new partners after their handshake. Because the card cleanup above sends no OCPI push, ask your CPO partners to pull Tokens once after the upgrade, so they see the deactivated cards as not valid.
- AWS CDK: the load balancer idle timeout default is now 600 seconds (was 120). It must stay above the longest CSMS or station ping interval plus 30 seconds. If you set
alb.idleTimeoutSecondsyourself, raise it to at least 270 for stations on the Cellular plan. See AWS. - Proxies. The API no longer reads
X-Forwarded-Protofrom an address outsideTRUST_PROXY. Behind any HTTPS proxy withoutTRUST_PROXY, sign-in cookies lose theSecureflag and the__Host-prefix, and the portal stream cap (10 per client) and the guest checks count every driver as the proxy address. SetTRUST_PROXYto your proxy addresses. Helm installs behind a plain HTTP gateway with an emptyapi.env.trustProxystill render, with a warning that every client shares one rate limit: set it anyway. See Client Addresses. - OCPP connection limits. Each address may have at most 50 station connections waiting for their password check (
OCPP_MAX_PENDING_AUTH_PER_IP, default 50), and 20 wrong station passwords a minute lock out that address. Stations whose password succeeded within the last 24 hours connect at once. Above the limit for one station id, each address gets one password check a minute for that station, so a station without a recent success still connects within about a minute. Forwarded addresses count only fromOCPP_TRUSTED_PROXY_CIDRS: behind a proxy or load balancer without it, every station counts as the proxy address (the server logs a warning). SetOCPP_TRUSTED_PROXY_CIDRSto the proxy addresses. API rate limits now hold across replicas through Redis. See Connection Limits. - reCAPTCHA checks the host and action. A CSMS or portal opened on a host other than those of
CSMS_URLandPORTAL_URL(a LAN address, an alias) gets 403RECAPTCHA_FAILEDwhile reCAPTCHA is on. Set both URLs to the hosts users open. See reCAPTCHA. - SSO accounts created before the upgrade. Until each one signs in by SSO once, the first SSO sign-in that asserts its email links it. Set
sso.allowedDomainsbefore you upgrade, or have SSO users sign in soon after. - Payment settings permission. The upgrade grants
settings.payment:readandsettings.payment:writeto every user with the admin role. Other users and API keys that manage payment providers need it granted. - Data retention values. The upgrade raises a stored audit, sign-in or station security event retention of 1 to 29 days to 30 and logs a notice. The worker reads such a value as 30 until then.
- Station data plans after the upgrade. The upgrade changes the heartbeat interval, the charging readings and the idle readings only where the setting still holds the earlier shipped value; a value you changed stays. Stations with a confirmed SIM run the Cellular plan, and every station gets its plan's values at its next boot. Helm writes
appSettings.ocpp.*on every deploy: an upgrade with--reuse-valueskeeps the earlier values (heartbeat 300) until you change them. See Data Usage Defaults. - Simulator target. The simulator
targetUrlis limited to the OCPP server URLs (OCPP_SERVER_URL,OCPP_TLS_SERVER_URL). - Avatar uploads need a CORS rule for the CSMS and the portal on self-managed S3 buckets. The CSMS and the driver portal upload avatar images straight to the S3 bucket with a presigned
POST. A self-managed bucket (Docker Compose, Helm, your own bucket) needs a CORS rule that allowsPOSTfrom both the CSMS origin and the portal origin. Without it, avatar uploads fail in the browser. Generated avatars and the mobile app are not affected. The AWS CDK bucket already allows both origins. See S3 Bucket. - Redis ACL for the avatar cleanup queue. The api Redis user adds jobs to the new
driver-avatar-cleanupworker queue and needs~bull:driver-avatar-cleanup:*. The Helm chart, the CDK stack and Docker Compose ship the rule. A Compose stack reads it when Redis restarts (scripts/docker-build.shrecreates Redis). If you run your own Redis or Valkey ACL, add the pattern to the api user. - AI insights scope after the upgrade. An install where at least one site had automatic insights switched off starts with the scope Selected Sites: every site that had insights on is selected, and the switched-off sites stay out. Every other install starts with All Sites. Review the AI Insights Scope card in Settings > AI > Automatic Insights. See Automatic Insights settings.
- Simulator file transfers to private hosts. Simulated stations now really upload log files and download firmware. Outside
NODE_ENVdevelopmentortest, they reach a loopback or private address only for a host listed inCSS_FILE_TRANSFER_ALLOWED_HOSTS(Helmcss.env.fileTransferAllowedHosts). An SFTP server needs its host key inCSS_SFTP_HOST_KEY_SHA256(Helmcss.env.sftpHostKeySha256) while stations verify server certificates. Docker Compose allows its ownftp,sftpandfile-receivertest servers. See Simulator Environment Variables. - OCPP processes of an earlier release in System Information. During a rolling upgrade, an OCPP process that still runs an earlier build shows "Not reported (the process runs an earlier release)" instead of its settings. It reports them after its restart.
- Database storage. Chaos mode and large networks fill
ocpp_message_logsandmeter_valuesquickly: a test network of about 2,000 simulated stations reached 90 GB. Plan the database disk, or lower OCPP Messages and Meter Values in Settings > Data Retention. See Data Retention. - Required MFA and passkeys. Required MFA is off by default, and passkey sign-in is off by default. Passkeys as a second factor are on by default. Passkeys need HTTPS or
localhost: a portal served on an IP address does not offer passkeys. - Choose
WEBAUTHN_RP_IDbefore users register passkeys. Changing it later makes every registered passkey unusable. See Environment Variables. - Scripts that turn MFA off. Users who disable MFA from a script, or who keep only a switched-off method, should set up a passkey or save recovery codes first.
- Password policy. The upgrade stores today's password rules (12 characters, uppercase, lowercase and a number). Driver app builds older than v0.1.48 keep their built-in rules and show a generic weak password message when the policy asks for more.
- SIM providers and the Redis ACL. The SIM provider and station signal switches are off by default. While both are off, nothing calls a provider and no station page or API reply changes. The api Redis user needs the
~sim:*key pattern. The Helm chart and the CDK app carry it. If you run Redis with your own ACL file, add~sim:*to theapiuser before you turn on a SIM provider. - Migration.
0479_release_0_1_48.sqlis one idempotent migration. Installs that ran a v0.1.48 prerelease run it whole again on the next migrate.
New
Sign-in, MFA and passwords
- Password policy in Settings > Security > Passwords and Sessions. Set the minimum length (8 to 64, default 12) and whether an uppercase letter, a lowercase letter and a number (on by default) and a special character (off by default) are required. One policy covers operators and drivers. It applies the next time someone sets a password; sign-in never checks it, so current passwords keep working.
- Session lifetimes in the same tab. Set the operator and driver sign-in check interval (5 to 1440 minutes, default 60) and inactivity timeout (1 to 90 days for operators, 1 to 365 for drivers, default 30). New values apply at sign-in and at the next background renewal.
- Live password checklists. The CSMS and driver portal password forms show the configured rules, and a rejected password names the rules it misses.
- API.
PUT /v1/security/password-policyandPUT /v1/security/sessions(settings.security:write, access to every site).GET /v1/security/publicreturnspasswordPolicy.GET /v1/security/settingsreturnspasswordPolicyandsessionLifetimes. A 400WEAK_PASSWORDreply carriesmissing(minLength,uppercase,lowercase,number,symbol,maxLength). - Helm and CDK.
appSettings.security.password.*andappSettings.security.session.*set the same values. They default to empty, so the value saved in the dashboard stays. - Required MFA (optional, off by default). An admin can require MFA for every operator or for chosen roles, and separately for drivers, in Settings > Security > MFA. Only when it is on do users without MFA set it up at their next sign-in, before the session starts, and renewals check the policy again. See Required MFA and Settings > Security > MFA.
- Recovery codes. Ten single-use codes per operator and driver, shown once when MFA is turned on, with a download option and a regenerate flow. A recovery code signs in when the usual factor is not at hand. See Recovery Codes.
- Passkeys and physical security keys. Operators and drivers can register passkeys and security keys as a second factor, and optionally sign in with a passkey alone (Settings > Security > MFA). Manage them in Profile in the CSMS and in Account > Security in the portal. See Passkeys and Security Keys.
- Security notices. Operators and drivers get an email (and a text for SMS sign-in accounts) when a factor is added or removed, MFA is turned on or off, an admin resets MFA, recovery codes are regenerated or used, the phone or email changes, or an API key is created. See Security Notices.
- Driver sign-in protection. Drivers get the operator sign-in lockout: 5 free failures, then a delay that doubles from 30 seconds up to 15 minutes. See Sign-In Lockout.
- Phone change confirmation. Changing the phone number needs a proof when the account holds a factor. Accounts that sign in with SMS codes confirm the new number with a code sent to it.
- API. New routes for recovery codes (
/v1/users/me/mfa/recovery-codes,/v1/portal/driver/mfa/recovery-codes), passkeys (/v1/users/me/webauthn/*,/v1/portal/driver/webauthn/*, passkey sign-in under/v1/auth/passkey/*and/v1/portal/auth/passkey/*, the passkey second factor under/v1/auth/webauthn/*and/v1/portal/auth/webauthn/*), factor proofs and phone verification.GET /v1/users/me/mfaandGET /v1/portal/driver/mfalist the accepted proof kinds. - Helm and CDK.
appSettings.security.mfaRequired,mfaRequiredRoles,mfaRequireDrivers,mfaWebauthnEnabledandmfaPasskeyLoginEnabled(CDKappSettings['security.mfa.*']) set the same values. Empty keeps the dashboard value. Both refuse required MFA while email and authenticator app codes are both off. See Helm chart values and AWS CDK. WEBAUTHN_RP_ID(API, optional) sets the passkey relying party ID. Unset, operators use the CSMS host and drivers the portal host. See Environment Variables.- Sign-in lockout keeps known devices signed in. A browser that signed in to an account before keeps its own failure count during a lockout, so failed attempts from elsewhere cannot lock out the owner on that browser. The lockout count resets after 24 hours without a failed attempt. See Sign-In Lockout.
- Page security headers. The CSMS and portal images send a Content Security Policy, frame, referrer and
nosniffheaders.
Stations, SIMs and data plans
- SIM providers (optional, off by default). Connect the Hologram account of your cellular stations in Settings > Integrations & Features > SIM Providers to see each station's SIM status, cellular data sessions and SIM alerts. While Hologram and Read Signal from Stations are both off, nothing calls a provider, no OCPP command is sent, and no station page or API reply changes. Only a person's click calls the provider: Refresh, Test Connection and Reset Network. Configuring the account needs access to every site. See SIM Providers and Settings > SIM Providers.
- Station Connectivity tab. Shows while a SIM provider or station signal reads are on, with the ICCID and IMSI field on the Details tab. It holds the SIM status, the data sessions, the five newest SIM alerts and a timeline of connects and disconnects. Needs
stations:read. See Station Connectivity.- Refresh fetches the SIM status, sessions and alerts for a period of 1 to 31 days or a custom range, and reads the station's signal in one OCPP request when signal reads are on. Opening the tab reads only the database. See Refresh.
- SIM attachments. A station's SIM is attached by its ICCID. An attachment from the ICCID the station reported, or one a site-restricted user entered, fetches nothing until a user with access to every site confirms it with Confirm SIM. See SIM Attachment.
- Signal grade. The modem signal the station reports is graded good, fair or poor (at or above -65 dBm good, at or above -85 dBm fair, both thresholds are settings). See Signal Grade.
- Offline causes. Each disconnect on the timeline is labeled with its likely cause (CSMS restart, SIM account, station, network, pending or unknown) from stored data only. See Likely Cause of a Disconnect.
- Reset Network asks Hologram to reset the SIM's cellular connection. It needs
stations:writeand a confirmed match, and allows one reset a minute per station. See Reset Network.
- Low signal monitor (optional, off by default). With station signal reads on and a low signal alert threshold set, each OCPP 2.0.1 and 2.1 station that supports monitoring gets a
LowerThresholdmonitor on its modem signal. Low signal and restored events show on the timeline. Clearing the threshold or turning signal reads off removes the monitors. See Low Signal Alert. - API. New routes
GET /v1/stations/:id/sim,GET /v1/stations/:id/sim/timeline,POST /v1/stations/:id/sim/refresh,POST /v1/stations/:id/sim/attachment/confirm,DELETE /v1/stations/:id/sim/attachment,POST /v1/stations/:id/sim/network-reset,GETandPUT /v1/settings/sim-providersandPOST /v1/settings/sim-providers/:provider/test, with newSIM_*error codes.PATCH /v1/stations/:idacceptsiccidandimsi. See Station Connectivity error codes and SIM Providers error codes. - Helm and CDK.
appSettings.sim.*(CDKappSettings['sim.*']) set the SIM provider and station signal settings. Empty keeps the dashboard value. The Hologram API key is entered in the dashboard only. - Station data plans. Two sets of OCPP station settings: Wi-Fi or Wired (the default) and Cellular, for stations on cellular data. Each plan has its own heartbeat interval, CSMS ping interval, idle readings, end readings and charging readings, with separate values for AC and DC stations. A station is DC when a connector of an in-service EVSE is CCS1, CCS2, CHAdeMO, or GBT, Tesla or NACS above 22 kW. The CSMS sends the plan's values to each station at every boot, and at once to online stations when their plan or charging kind changes. Edit the values in Settings > Integrations & Features > OCPP, on the Wi-Fi or Wired and Cellular tabs. See Data Usage Defaults and Cellular Stations.
- Site and station plans. A station uses, in this order, the plan set on the station, Cellular when a SIM attachment is confirmed on it, the site's plan, and Wi-Fi or Wired. The Data Plan card on the station Configurations tab shows the plan, where it came from and AC or DC, and sets or clears the station's choice (
stations:write). The Data Plan card on the site Details tab sets the site's plan and names how many stations change before you save (sites:write). See Pick the Cellular Plan and Site Data Plan. - Four configuration templates. "Wi-Fi or Wired Data Plan (OCPP 1.6)", "Cellular Data Plan (OCPP 1.6)", "Wi-Fi or Wired Data Plan (OCPP 2.0.1 and 2.1)" and "Cellular Data Plan (OCPP 2.0.1 and 2.1)" carry the station-side settings each plan needs, and target their plan's stations through the new Data Plan target filter. Template variables take a variable instance (OCPP 2.0.1 and 2.1), and the values the data plan sends never count as template drift. See Which Template to Use.
- WebSocket compression. The OCPP server negotiates permessage-deflate on the plain and TLS listeners.
OCPP_WS_COMPRESSION=false(Helmocpp.env.wsCompression) turns it off. Stations that do not offer it send uncompressed JSON as before. See WebSocket Compression. - Idle readings off. An idle reading interval of 0 is now sent to the station and turns idle readings off. The Cellular plan ships with 0.
- API and Helm.
PATCH /v1/stations/:idandPATCH /v1/sites/:idtakedataPlan(wifi_or_wired,cellularornull), andGET /v1/sites/:id/data-plan-impact?dataPlan=counts the stations that would change plan. New settingsocpp.csmsPingInterval,ocpp.dc.*andocpp.cellular.*. HelmappSettings.ocpp.csmsPingInterval,appSettings.ocpp.dc.*andappSettings.ocpp.cellular.*(empty keeps the dashboard value). Connector typeCCS1is added.
- Site and station plans. A station uses, in this order, the plan set on the station, Cellular when a SIM attachment is confirmed on it, the site's plan, and Wi-Fi or Wired. The Data Plan card on the station Configurations tab shows the plan, where it came from and AC or DC, and sets or clears the station's choice (
- Data Retention settings tab. Settings > Data Retention edits every retention setting in one place: operational logs, security and audit history, sign-in sessions and tokens, AI conversations and insights, and SIM data, each with its range and default. It needs access to every site, and the fields are read-only without
settings.system:write.GETandPUT /v1/settings/data-retentionoffer the same. A connection log retention of 0 keeps it forever. See Data Retention and Settings > Data Retention. - Session meter summary. Each charging session keeps its peak power and its last state of charge, written with each meter reading, so deleting old meter values no longer changes them. The NEVI report reads the session's peak power, and the portal session detail reads the battery level from the session. The upgrade fills both from the meter values still stored. See Sessions After Their Meter Values Are Deleted.
- SIM inventory (with the SIM provider feature, off by default). A SIM list (
/sims) and a SIM page with Details, Sessions, Attachments and History tabs. Add SIMs one at a time, by CSV import (with a template) or by entering an ICCID on a station, and export the list with the filters on screen. A SIM's sessions, alerts and status belong to the SIM, and each station reads them through its current confirmed attachment. Users with access to every site attach, confirm, detach, deactivate, reactivate and delete SIMs. A site-restricted user sees only SIMs with a confirmed attachment to a station of their sites. New permissionssims:readandsims:write(the default operator role gets both, the viewer role read). The page shows only while a SIM provider or station signal reads are on. See SIMs. - OCPI approved roles. Each partner page has an Approved Roles card that shows the roles the partner declared and holds, and checkboxes to approve them. See Partner Registration.
- New error codes.
DATA_PLAN_SETTING_IN_TEMPLATE,VEHICLE_ID_IN_USE,AUTOCHARGE_CLAIM_REQUIRED,PREPAID_BALANCE_CONFLICT,SESSION_REQUIRED,PERMISSIONS_EXCEED_OWN,PAYMENT_PROVIDER_NOT_CONFIGURED,MANUAL_SIGNING_REQUIRED, the SIM inventory codes (SIM_ATTACHED,SIM_ATTACHED_ELSEWHERE,STATION_HAS_SIM,SIM_ATTACHMENT_CHANGED,SIM_ATTACHMENT_NOT_CONFIRMED,SIM_INACTIVE), and the framework codesUNAUTHORIZED,NOT_FOUND,PAYLOAD_TOO_LARGEandUNSUPPORTED_MEDIA_TYPE.
Dashboard, avatars, AI and logs
- Customizable dashboard charts. Each user arranges the charts of the Live dashboard: the Customize button (a sliders icon next to the mode toggle) shows a drag handle and a hide button on every chart, hidden charts wait under Hidden, and Reset Layout returns to the default. Drag and drop works with the mouse, touch and the keyboard. The stat cards keep their fixed order and rows, and on phones the financial cards scroll sideways. The layout is saved on the server per user (
GET,PUTandDELETE /v1/users/me/dashboard-layout). Historical and Trend keep the default order. A layout never shows a chart the user's permissions do not allow. See Customize the Live Dashboard. - Four new dashboard charts, hidden by default. Locations by City (accepted stations per city), EVSEs by Connector Type, Plug & Charge (sessions and energy per day, and the active contracts for users with access to every site and
drivers:read) and Free Vend (sessions and energy per day, and the sites with free vend on). New routesGET /v1/dashboard/locations-by-city,/evses-by-connector-type,/plug-and-chargeand/free-vend. See Charts. - Avatars for drivers and CSMS users. Everyone edits their own avatar: a generated avatar (with Random, and None for hair or headwear, glasses and facial hair) or an uploaded JPEG, PNG or WebP image smaller than 1 MB, cropped to a centered square, scaled to at most 512 x 512 pixels and re-encoded without metadata. One avatar at a time. Drivers use Edit Avatar on Account > Personal Info in the portal (the registration form offers the generator only) and the avatar screen under Account in the mobile app; upload shows only when file storage is set up. CSMS users use Profile > Personal Info, and the sidebar user area shows the signed-in user's avatar. Operators with
drivers:writeset or remove a driver's avatar in the driver edit form and on the New Driver page, and operators withusers:writeset or remove another user's avatar in the user edit form and on the New User page (within their site scope). The CSMS shows avatars on the Drivers and Users lists, the fleet Drivers tab and the driver and user pages. Every change is recorded in the account's history. Erasing a driver or deleting a user deletes the avatar files. New routes:GET,PUTandDELETE /v1/portal/driver/avatar,POST /v1/portal/driver/avatar/upload-urland/image;PUTandDELETE /v1/drivers/:id/avatar,/v1/users/me/avatarand/v1/users/:id/avatar, each withPOST .../upload-urland.../image. Driver and user replies carryavatar,POST /v1/portal/auth/registertakes an optional generatedavatar, andGET /v1/portal/featurescarriesavatarUploadEnabled. See Driver Avatar, User Avatars, Profile Avatar, Avatar in the portal and Avatar in the mobile app. - AI insights scope. Settings > AI > Automatic Insights has an AI Insights Scope card: All Sites (default) or Selected Sites, with a switch per site in a searchable list that scrolls inside the card and a switch per station inside a selected site. Sites and stations outside the scope get no insights or background worker suggestions, and their open insights are dismissed and waiting actions expire when they leave it. Station and site details show the AI Insights status with the reason when insights are off, and Suppress is offered only in the scope (409
AI_INSIGHTS_NOT_IN_SCOPEotherwise). The insight, scope and background worker settings are disabled while Turn on system-initiated AI is off. The site AI Usage card no longer has the insights switch. New routesGETandPUT /v1/settings/ai/insights-scopeandGET /v1/settings/ai/insights-scope/sites/:siteId/stations;GET /v1/stations/:idandGET /v1/sites/:idreturnaiInsightsScope. See Automatic Insights settings. - Real log uploads in the simulator. Simulated stations answer GetLog and GetDiagnostics by uploading a generated log file to the location over FTP, FTPS, SFTP, HTTP or HTTPS, and report Uploading, then Uploaded or UploadFailure, with the request's retries. Firmware downloads use the same protocols. Docker Compose adds
ftp,sftpandfile-receivertest servers under thetoolsprofile. See Simulator Actions. - More demo data.
SEED_DEMOadds Plug and Charge, free vend and SIM demo data, so the new dashboard charts and the SIM pages show data on a demo stack. - Audit entry details. A click on a row of the Audit Log page or any History tab opens the entry: the action, entity, time, actor, AI channel and notes, and a Changes table with Before and After per field (Show Unchanged Fields lists all). Go to Entity opens the entity's page when it exists and you may open it. The Audit Log filters stay in the page URL. See Entry Details.
- Date filters on the log pages. From and To day filters on every Access Logs tab and on the Audit Log page, kept in the page URL, with Clear Dates.
GET /v1/access-logsandGET /v1/worker-logstake optionalfromandto. See Access Logs. - System Information search and new sections. The dialog has Search settings and a Section filter, and adds the Security, OCPP Processes (each OCPP process with its own settings), SIM Providers and Station Data Plans sections, the station TLS address, the number of simulator stations, and the webhook signing secret and Hologram key flags. See System Information.
- Website. The evtivity.com home page scrolls smoothly and reveals its sections as you scroll (not under reduced motion), and the site uses one type scale with Inter at its display optical size for large headings. See evtivity.com.
Changed
- Company Info images are replaced on upload. The logo, QR code icon, favicon and social share image have no Remove button. Uploading a new image replaces the current one.
DELETE /v1/settings/:keystill clears a value. See Company Info. - Bulk reservations take one station in several slots. A station can be in several slots of a bulk reservation, one connector each, and a connector chosen in another slot is not offered again. Two bulk reservations of a fleet can have the same name. See Create a Bulk Reservation.
- The operator sign-in page links to the "Charging Portal" without the company name in front.
- The AI assistant panel loads ahead, so its first open is instant.
- A station log upload keeps one row per upload in the station's log upload history, updated by each status report, and stores the OCPP 2.1
statusInfoof a failure.
Changed defaults
- Heartbeat interval 300 to 900 seconds on the Wi-Fi or Wired plan (3600 on Cellular). Helm
appSettings.ocpp.heartbeatIntervaldefault 900. - CSMS ping. Every 30 seconds to every station becomes every 120 seconds (Wi-Fi or Wired) or 240 seconds (Cellular), and only after that long without any frame from the station. A station that answered a ping is disconnected after 3 times its plan's ping interval without a frame (6 minutes Wi-Fi or Wired, 12 minutes Cellular).
- Idle and offline timeouts follow the plan's heartbeat: Wi-Fi or Wired 30 minutes idle and 45 minutes heartbeat timeout (were 10 and 15 minutes), Cellular 2 and 3 hours.
- Idle readings send only
Energy.Active.Import.Register(was five values). The seed default idle interval is 900 seconds (was 60 inseed.config.default.json). Cellular sends no idle readings. - Charging readings on AC stations drop
VoltageandSoC:Energy.Active.Import.Register,Power.Active.Import,Current.Importevery 60 seconds. DC stations addSoCand read every 30 seconds. Cellular reads AC every 300 seconds and DC every 60 seconds. - WebSocket compression on (
OCPP_WS_COMPRESSION=true). - CDK
alb.idleTimeoutSeconds120 to 600. - Helm
css.enabledtruetofalse(see Breaking changes).
Fixes
Payments and charging
- A guest checkout without a payment hold can no longer charge at a paid tariff.
- A selected payment provider the process cannot use (no credentials, or the simulated provider with
PAYMENTS_ALLOW_SIMULATED=false) no longer counts as payments off: a paid session stops unbilled at start, and the portal start answers 400PAYMENT_PROVIDER_NOT_CONFIGURED. - A driver whose saved card belongs to a provider that is no longer configured is stopped at start and asked to add a card, while any payment provider is active.
- Saving a card form no longer overwrites a prepaid balance with a stale value.
- Hold and capture amounts are bounded to 1 to 1,000,000 cents, the operator override included, and a capture is checked against its hold.
- A site-restricted user's payment config save keeps the platform fee and is refused only when it changes the fee.
- A payment webhook event is no longer lost when the process stops between recording and applying it.
- The Stripe Connect endpoint no longer accepts payment events, an Adyen amount must match the record's currency, and the worker no longer takes the payment provider from the job data.
- An operator capture reads the platform fee from the session's site, not the station's current site.
- A guest link no longer stops a session after it ended, and crediting a paid invoice no longer restores the fleet's credit exposure.
- Session invoice creation is for users with access to every site only.
- Portal forgot-password sends a link only to a verified email address.
Sign-in and accounts
- A driver password change, password reset, sign-out, activation, deactivation or erasure now ends access tokens issued before it at once, not only the refresh tokens. A password change keeps the device that made it signed in.
- The web portal refresh cookie lived 7 days while its stored session lived 30. Both now follow the driver inactivity timeout.
- New passwords are at most 128 characters. Sign-in and current password fields stay uncapped.
- A new password shorter than the policy minimum but at least 8 characters now answers 400
WEAK_PASSWORDwithmissing: ["minLength"]instead ofVALIDATION_ERROR. Under 8 or over 128 characters still answersVALIDATION_ERROR. - Saving the reCAPTCHA, MFA, password or session settings writes all of its keys in one transaction, so a failed save changes nothing.
- The old portal
/profilepage now opens Account > Security. - The station OCPP command dialogs keep a long JSON response inside the dialog on phones.
- Passkey autofill on the CSMS and portal sign-in pages no longer restarts without end after a refused browser prompt.
- The next sign-in in a tab after Log Out opens the dashboard, not the page that was open at Log Out, whoever signs in.
- SSO sign-ins are bound to the browser that started them, and the SSO binding cookie uses the
__Host-prefix over HTTPS. - Percent-encoded paths (such as
/v1/%70ortal/...) are refused with 400 before any route hook runs. - Account, sign-in and admin user and driver replies send
Cache-Control: no-store. - MFA codes are bound to their purpose and, for SMS, to the number they went to. Code sends are capped per account and at 3 an hour per number across accounts.
- Parallel wrong MFA guesses can no longer race past the lock, and a settings read failure no longer lets a switched-off method through while another factor exists.
- A driver portal cookie resent as a Bearer token no longer passes as the mobile app.
- Vite is updated to 8.3.5 for the launch-editor advisories (GHSA-r825-w8mm-5pxw and related). This affects the development servers only.
- Password sign-in and MFA verification no longer accept form posts. Every route except the SAML callback answers 415 to
application/x-www-form-urlencodedbodies. - Refresh tokens are checked by type and by operator or driver realm.
- Forgot-password, sign-in and registration take the same time for known and unknown emails.
- An inactive operator with the right password gets 401
INVALID_CREDENTIALS, the same answer as a wrong password. - The SSO email domain check reads the whole address, and SAML replies are no longer stored in the access log.
- The operator password reset email goes to the stored address, not the typed one.
- The event streams no longer accept a token in the query string, and the unsigned
csms_tokencookie is gone. - reCAPTCHA checks the host and the action of each answer.
- Signing out of the CSMS clears the data cached in the browser.
- User history rows keep the earlier sites and permissions, and setting history rows no longer store encrypted secrets.
- Creating a user runs in one transaction.
Permissions and site access
- Station history and metrics follow the site a session ran at, not the station's current site.
- Tariff routes check that the tariff belongs to the pricing group in the path.
- The CSMS event stream sends only the event families the user may read.
- A report generated by a user with no sites is no longer shown to every site-restricted user.
- An admin can no longer set their own password through the admin reset route.
- The assistant can no longer set outbound data destinations, write notification templates, or propose email and phone changes for drivers, sites and NEVI operators. Its confirmation card shows masked argument values, and a confirmed command is checked again against the station's OCPP version.
Stations, OCPP and Plug and Charge
- Plug and Charge certificate chains no longer accept an end-entity certificate as an issuer, and a contract certificate must belong to the eMAID it authorizes.
- OCSP verification accepts a responder only through a chain to an operator or local CA, and OCSP, CRL and webhook error replies are read with a size cap.
- The security profile 3 client certificate is matched by fingerprint, not by serial number alone.
- A pending security profile is promoted only when it is an upgrade.
- Incoming station messages are validated before they are logged or queued, with a size bound.
- Failed connections are counted per station in the connection log instead of one row each.
- The
Basicauthorization scheme is compared without case. - The simulator connects only to the configured OCPP server URLs.
- Sessions keep their peak power and last state of charge. The NEVI report no longer reads W values as kW and no longer adds inlet and per-phase readings into one peak.
OCPI and integrations
- OCPI
GET /emsp/tokensrefuses partners without the CPO or HUB role. - Two OCPI credentials updates at once no longer leave two active tokens.
- The Hubject token URL and the AI base URL go through the private address check, also at send time, and the address check reads the 64:ff9b:1::/48 range and the AWS IPv6 metadata address correctly.
- Station images are size-checked, content-checked and re-encoded, and SVG files are no longer served inline.
- Email previews no longer load remote content, XLSX exports no longer show the protecting quote, CSV exports protect formulas after
;and TAB, and presigned downloads open withnoopener. - The assistant's document fence holds against variants of the closing tag.
Deployment and platform
- Updated DOMPurify and Fastify for the dependency audit advisories.
CORS_ORIGIN=localparses addresses instead of trusting host names with a private prefix.- Cookies get the
Secureflag only from a trusted proxy, and process logs redact everytoken=in a URL. - Images no longer copy local secret files from the build context, and base images use pinned tags.
- Sentry starts without personal data collection.
- Framework replies use their own codes (
UNAUTHORIZED,NOT_FOUND,PAYLOAD_TOO_LARGE,UNSUPPORTED_MEDIA_TYPE) instead ofVALIDATION_ERROR. GET /v1/portal/brandingalways returnsportalUrl, falling back toPORTAL_URL.- The API and worker images build again with the SIM provider mock loaded at run time.
CSMS and portal
- SIM pages, the station Connectivity tab and list addresses stay in step, and a site user's SIM history rows outside their sites' periods stay hidden.
- The MFA Verify button stays enabled and explains an incomplete code on submit.
- Passkeys that cannot be used say why, and verification uses the sign-in layout.
- The new security and SIM screens use the shared form patterns. The API key edit checks on submit, and the quick start uses the shared tabs.
- The German portal addresses drivers with "du" everywhere.
Other fixes
-
An OCPP 2.0.1 and 2.1 NotifyEvent item with
cleared: true(a monitored value back to normal) no longer opens an event alert and no longer changes an open one. The operator still acknowledges the open alert. -
An OCPP 2.1 TransactionEvent the station replays from its offline queue (
offline: true) no longer changes the live connector status. Before, its timestamp could hide the station's newer StatusNotification after a reconnect and leave a connector shown as idle while the station reported it available. The session, meter values and costs of the replayed event are recorded as before. -
The simulator lists the OCPP MessageTimeout setting once, with its Default instance, as OCPP 2.1 defines it.
-
The Sites list shows the search and all three filters on one row.
-
Dependency security updates: zod 4 (fixes a high advisory where a large array in a request could exhaust server memory during validation), plus undici, fast-uri and nodemailer. Request and response shapes are unchanged.
-
Number settings and internal checks validated by zod now refuse Infinity and integers beyond the safe range.
-
The bulk reservation dialog shows No Connectors for a station without connectors, and an error with Retry when the connectors fail to load, instead of loading without end. See Create a Bulk Reservation.
-
Back to Sign In is centered on the sign-in verification step.
-
The Add Holiday and Bulk Add holiday dialogs use the spacing of the standard dialog and describe what they do.
-
Security profile 0 reads No Auth for OCPP 1.6 stations, as for OCPP 2.0.1 and 2.1.
-
Passwords in file transfer URLs are masked wherever they are stored. A firmware, log upload or diagnostics URL can carry a user name and password (
ftp://user:password@host/file). The station still gets the full URL, but the OCPP message log, domain events, access log, firmware update and log upload history, audit rows, AI tool results and queued commands that left the queue show the password as[redacted], and the upgrade masks the copies stored before. Firmware campaigns show the full URL only to users withsettings.system:read. A URL password with a space, a control character, or a raw/,?or#answers 400VALIDATION_ERROR(percent-encode those characters), and so does sending the masked password back. See Firmware Updates. -
File transfer URLs are limited to 2,000 characters (the OCPP 2.1 limit) on the OCPP command routes and firmware campaigns (400
VALIDATION_ERROR). See Firmware Updates. -
The simulator verifies SFTP host keys while it verifies server certificates and never logs a transfer URL password.
-
Phone layouts: detail page headers, the tab strip (which scrolls the active tab into view), station, site and report rows, Settings Conformance and Content, the CA Certificates actions, the System Information dialog and the portal Popular Times days fit a phone screen.
-
The dashboard Customize tooltip opens to the left, the EVSEs by Connector Type chart shows full connector type labels, and the audit entry Changes table wraps long values.
-
The support case sidebar labels have space above their dropdowns, and Prev and Next sit at the right of the driver page header.
-
Settings > S3 Bucket shows the CORS rule for
POSTuploads from both the CSMS and the portal origins. -
Charging station conformance runs (OCPP 1.6, 2.0.1 and 2.1) download firmware from the test file servers the runner starts, so the firmware download tests run to the end. See Conformance.
-
Rows of buttons on detail pages, dialogs and settings wrap at phone width, and long button labels wrap inside the button.
Mobile app
The v0.1.48 driver app ships with this release:
- Framework error codes (
UNAUTHORIZED,NOT_FOUND,PAYLOAD_TOO_LARGE,UNSUPPORTED_MEDIA_TYPE) have their own messages, and a 429 says "Too many requests. Wait a moment and try again." - Field errors and portal messages show in the driver's language: "Check the highlighted fields." when the reply names fields.
- The app registers its push token again after every new session, password change and MFA change, and keeps the session on a temporary renewal error (408, 429, 5xx).
- Turning MFA off, adding a factor and changing the phone number when a proof is needed open the driver portal, with the reason in the driver's language.
- Turning MFA on shows the recovery codes once, with Copy Codes and I Saved Them.
- Adding a card that is another driver's vehicle id, turning Autocharge on before the vehicle was seen in a session, and a start without a usable payment method (400
PAYMENT_METHOD_REQUIRED, which reloads the cards and offers to add one) each explain what to do. The card form no longer offers the VIN, EVCCID and MAC address types. - Email and SMS MFA codes send their challenge id.
- German texts address the driver with "du".
- Avatar screen under Account. Tap the avatar at the top of the Account hub to create a generated avatar (with Random, and None for hair or headwear, glasses and facial hair) or upload a photo from the library, cropped to a square in the picker. JPEG, PNG or WebP smaller than 1 MB. Upload shows only when the operator has set up file storage. Remove Avatar shows the initials again. See Avatar.
- Generated avatars render on the device and are checked against the same option list as the server. An uploaded photo is never cached on the device. A photo whose file size cannot be read shows a message instead of failing.
- The app asks for photo library access with its own permission text (
expo-image-picker). It adds DiceBear (@dicebear/core10.7.0,@dicebear/styles10.6.0). The Avataaars design by Pablo Stanley is credited in the third-party notices.