Tools and reading for people who run a fiber network.

What changes

• “About OLT Controller”: a button in the header shows:
  • the installed version;
  • system details, supported vendors and technology;
  • your session;
  • the status of your legal acceptances;
  • “Copy details for support”.
• Legal documents: Terms and Conditions, Privacy Policy, Cookie Policy and Intellectual Property Policy, at /legal/….
  • They are public: you can read them without signing in.
  • They are versioned, can be printed or saved as PDF, and are linked in the panel footer.
• Acceptance is recorded:
  • a notice next to the “Sign in” button states that signing in and using the system means accepting them;
  • when you sign in, by password or passkey, and as you keep using it, the system records who accepted, which exact version (SHA-256 fingerprint), from which IP and with which browser;
  • that record cannot be edited or deleted.
• Suggestions and complaints (Help): all three roles can propose improvements or report bugs.
  • For a bug, you state how much it affects you and how to reproduce it.
  • The screen, version and browser are attached automatically.
  • eCloud replies and updates the status (New, Under review, Planned, In progress, Resolved or Dismissed), and the operator follows it in their panel.
• Syslog: reception parses the RFC 3164 and RFC 5424 formats, per message.

Before updating

• Includes migrations 068 (legal acceptances) and 069 (suggestions and complaints): the agent takes a backup and applies them. They only add tables.
• The legal documents ship as drafts: the company details are missing (tax ID, address, country, email addresses, jurisdiction, governing law, website). You should fill them in and have a lawyer review them.
• After updating, each user accepts the documents the next time they sign in, or at the next renewal of their session if they were already signed in.
• Restart backend and frontend: a few seconds without the panel.

What changes

• OLT syslog (optional, off by default). OLTs can send their log over syslog; syslog-ng receives it.
  • It stores one file per OLT per day, for bulk searches.
  • The system matches each message to its OLT and shows it on the OLT's detail page, with its severity, filters and search.
  • The source IP is assigned to an OLT, and each IP belongs to a single OLT across the whole platform: no one receives another party's device log.
  • Anything that arrives unassigned is shown in System settings › OLT syslog.
• Explore SNMP in Diagnostic tools. Walks a branch of the OLT's SNMP tree and shows what it returns, read-only, with row and time limits. Useful to see what each device returns before reading it in monitoring.
• Knowledge base:
  • new Huawei (20 articles) and C-Data (17) categories;
  • HALNy grouped into a single category with subcategories, like V-SOL and ZTE; old links keep working.
• Updates: each update's details show the changes of every version it applied, formatted.

Before updating

• Includes migration 070: the agent takes a backup and applies it. It only adds tables and a view.
• Restart the stack: a few seconds without the panel.
• Syslog stays off. To turn it on you need variables in the stack (DEPLOY §11.77) and to configure the OLT to send it.

What changes

• Forecasts. The system calculates when something that is slowly getting worse will cross its alert threshold, looking at the last 60 days. It forecasts four things:
  • the signal of each ONU;
  • the laser of each PON port;
  • the utilization of each uplink and PON port, against 80 % of its speed;
  • the temperature of each OLT.
• Where you see it:
  • in Monitoring › Forecasts, most urgent first, with when each item crosses and with what confidence;
  • in a new Grafana dashboard, “Forecasts”.
• How it's calculated: the trend uses a robust slope, so an odd day, like a fiber cut, doesn't announce a crossing that isn't coming. If the trend is doubtful, no date is stated.

Before updating

• Includes migration 067: the agent takes a backup and applies it. It only adds a table and a view.
• Restart the stack: a few seconds without the panel.
• Afterwards:
  • the first calculation runs after 3 minutes, and then every 6 hours;
  • it needs at least 7 days of data within 10, so on a new installation, or with little history, you'll see “no data” at first.

What changes

• GenieACS no longer runs out of memory. Each GenieACS service started one process per CPU: on a 4-CPU server that was 20 Node processes, and the system killed some of them even though the container showed as running. Now there are 9: two to serve the CPEs and one for each of the others.
• Backups complete fully again. The database and GenieACS were being backed up, but the files step failed when reading the VPN server's private key, and the backup container stayed red. Now the backup group can read that key, and only that group. It's the key that saves you from reconfiguring every router after a restore.
• NOC Grafana behind a proxy password. If Grafana's public address asks for a username and password at the proxy, the status page shows it as “responding, protected” and links to the dashboards; before, it flagged it as a problem.

Before updating

• No migrations.
• Restart the stack: a few seconds without the panel. GenieACS and the VPN restart; VPN tunnels are not dropped.
• After installing: System settings › Backups › “Back up now”. When it finishes, the backup container turns green.

What changes

• Optical monitoring alerts. Two new alerts:
  • low upstream: what the OLT receives from an ONU, compared against the module class of its port (C+, B+);
  • worn optical module: the laser current or temperature of a PON port close to its maximum.

The margins are in Settings › Alert thresholds.

• Root cause v2.
  • If a VPN tunnel goes down, the alert is for the tunnel, with its OLTs grouped under it, instead of one alert per OLT.
  • If an OLT goes down, whatever was already open for its ONUs is grouped under it.
  • You get an alert when an uplink in use goes down.
  • The Alarms screen shows the root with its dependents collapsed.
• OLT time.
  • The NTP suggestion for ZTE now includes ntp client; without that command, the C320 doesn't query any server.
  • The diagnostic says so, and recognizes when the difference is the time zone and not a wrong clock.
• Operator time zone, in My account. The admin picks it; everyone sees it.
  • “Time and NTP” compares the OLT's time against that zone and shows the system time in it.
  • “Manual time” defaults to the zone the OLT reports. Before, with an OLT on UTC−4 labeled “UTC”, it left the OLT four hours ahead.
  • The dashboard groups onboardings by day in that zone.
• System settings › NOC Grafana (platform). Shows, step by step, what's still missing to turn on Grafana:
  • the database role;
  • the container;
  • the public address;
  • the summaries.

When everything responds, it links to the four dashboards.

Before updating

• No migrations.
• Restarts the backend and frontend: a few seconds without the panel.
• If you're going to turn on Grafana: first set its variables in the stack and redeploy (DEPLOY §11.67); then install this version. During the migration, the agent sets the password for Grafana's read-only role.
• Afterward, each admin can set their operator's time zone in My account.

What changes

• OLT alarm logs, saved. Every hour the system reads the alarm log of each ZTE OLT (loss of signal, dying gasp, degraded signal, new ONUs) and saves it with the real time, even if the OLT's clock is wrong. A C320 keeps only the last 2,000 alarms, and they get overwritten in about 50 hours; now the history is kept.
  • On each ONU's detail page: “OLT alarms · 7 days”, with a plain-language reading (“It dropped 41 times: the link goes down and comes back”).
  • A new ONU that the OLT detects along with its version leaves the firmware on its detail page.
• Expanded optical monitoring. In addition to what each ONU receives, ZTE OLTs now measure:
  • the upstream, what the OLT receives from each ONU (a dirty connector usually shows up here first);
  • the power each ONU transmits;
  • the optical module of each PON port: power, laser current (bias) and temperature. The bias is compared with the maximum the module itself declares, and the OLT's detail page warns you when it's time to have a spare on hand.
• Grafana for the operations center (optional, off). Four dashboards: OLTs, optical network, traffic with 95th percentile, and OLT alarms. It's turned on from the stack; see DEPLOY §11.67.
• Hourly summaries of power, health, availability and traffic, to view months and years without going through every sample. No data is deleted.

Before updating

• Includes migrations 064, 065 and 066: the agent takes a backup and applies them. They only add tables, columns and views.
• Restart the backend and frontend: a few seconds without the panel.
• After restarting, the upstream and optical modules appear within about 10 minutes, and the first alarm log is read after 3 minutes (with the ~50 hours the OLT keeps).
• Grafana stays off. To turn it on, you need to update the stack in Portainer with its variables (DEPLOY §11.67).

What changes

• OLT time, date and NTP. New tool in OLTs › Tools and on each OLT's detail page (“Time and NTP” button).
  • Shows the device's time against the system's, whether NTP is syncing and what each server replies.
  • The administrator can set an NTP server or set the time manually, with the commands shown before sending.
  • On ZTE devices, the reading works out of the box. Both corrections are first set up in Edit OLT › Operations on the OLT, where the form offers the suggestion.
• Goodbye to PON ports the OLT doesn't have. A registered port with no ONUs that the OLT doesn't report (like the C320's “1/2”) is removed automatically during the optical reading. A port with ONUs is never touched.
• Why each ONU went down. With ZTE, the alert and the detail page say “no power (DyingGasp)” or “no optical signal (LOS)”. If most of a port or a NAP reported “no power”, the alert points to a power outage in the area rather than the fiber.
• The OLT version appears on its detail page. On ZTE C320 it is read with show version-running, because show version is ambiguous.
• CLI Explorer: a value like “Config state: fail” in a query is no longer shown as a command error.

Before updating

• Includes migration 063: the agent takes a backup and applies it. It only adds two columns to ONUs.
• Restart backend and frontend: a few seconds without the panel.
• The first optical reading, about 10 minutes after restarting, removes the nonexistent ports and starts saving the cause of each outage.

What changes

• Knowledge base: new ZTE category, with 27 articles on ZXA10 OLTs (C300, C320, C600) and ZTE ONUs, in four groups:
  • OLT access and management;
  • ONU onboarding (unauthorized, types, free ID, T-CONT and service-port, move or delete, restore, firmware);
  • diagnostics (status, optical power, SNMP, CPU and temperature, causes of outages);
  • ZTE ONTs.

They are mostly based on what was verified on a real C320.

• Knowledge base: new V-SOL category, with 15 articles on V-SOL OLTs and ONUs, based on their official material: access, GPON profiles, ONU onboarding, firmware, remote access and typical faults.
• HALNy ONTs in the ONU Models catalog: 19 GPON and XGS-PON models, with their official photo. The ONU detail page recognizes the HALNy brand by model or by the HALN serial prefix.

Before updating

• No migrations.
• Restart backend and frontend: a few seconds without the panel.

What changes

• Pagination across the entire panel. Every list lets you choose 15, 25, 50 or 100 rows per page; the default is 15. It applies to subscribers, ONUs, OLTs, alerts, logs, events, NAPs, zones, models, firmware, backups, TR-069, webhooks, integrations and platform. Changing a filter takes you back to the first page.
• OLT temperature in a ring, in Monitoring › Utilization, next to CPU and memory. It turns amber and red according to the configured alarm thresholds; with no data it stays empty instead of showing 0 °C.
• Node 24 LTS. The system images move from Node 20, which no longer receives security patches, to Node 24.

Before updating

• No migrations.
• If any integration uses the subscribers or ONUs API without specifying the page size, it now receives 15 results per page instead of 50. The maximum is still 200.
• Restart the backend and frontend: a few seconds without the panel.

What changes

• An OLT that's slow to negotiate SSH no longer shows as down. The system waits up to 20 seconds for the OLT to finish negotiating the SSH connection; it used to be 8. A ZTE C320 behind a VPN takes between 8.5 and 10 seconds and showed as “Offline” while the network was responding. This also caused the CLI explorer, the unauthorized ONU scan and onboarding to fail.

Before updating

• No migrations.
• Restart the backend and frontend: a few seconds without the panel.
• The OLT shows as online again on the first poll, about a minute after restarting.

What changes

• ONU onboarding on ZTE OLTs. The system asks the OLT which ONU numbers the port is using and adds the new one with the first free number. It also counts the numbers of ONUs that are powered off or authenticated by LOID. If it can't read the port, it doesn't send the onboarding. The number is saved with the ONU. To use it, the service plan needs its ZTE command template with {onu_id}.
• Measurements a person requests no longer wait. The installer's “Measure” and “Measure now” tell you right away if the OLT is busy with another operation, instead of waiting up to 25 seconds. The automatic reading every 10 minutes still waits its turn.
• Charts with the theme's colors and map legends that are readable in the light theme.

Before updating

• No migrations.
• Restarts the backend and frontend: a few seconds without the panel.
• On a ZTE, a plan change or a suspension from billing still resends the full onboarding, and the OLT rejects it because the ONU already exists. Nothing breaks, but for now onboarding is the only thing that works on ZTE.

What changes

• New Knowledge base in Help › Knowledge base: 96 articles in Spanish about GPON and ONTs, in 9 categories. They cover basic concepts, firmware upgrades over LAN and from the OLT, routers, provisioning (XML, TR-069, REST API), ONT configuration on OLTs from 12 vendors, device features, FAQs and web configuration.
• It has a search that ignores accents and capitalization and also searches inside the article text. Categories and searches have their own link, which you can bookmark or share.
• All three roles can use it (administrator, technical support and installer). It ships with the system and isn't edited from the panel.

Before updating

• No migrations.
• Restart backend and frontend: a few seconds without the panel.

What changes

• An ONU registered twice on the OLT is measured where it's working. If an ONU has an old registration on another port, the one the OLT reports as online now takes precedence, even if the inventory says otherwise. The backend log warns you when the inventory points to the old registration.
• “Location” on the ONU's detail page also corrects the PON port, not just the card and the number. It only changes the inventory: nothing is sent to the OLT.
• An SNMP that only half answers no longer causes false outages. If the OLT doesn't answer with the status of its interfaces, the status stays unknown and the ports aren't considered down. The write SNMP community test no longer decides whether the OLT is online.
• Fewer repeated alerts and emails. When an OLT comes back from an outage, only the alerts of the ONUs that went down with it are closed. The ones that were already off stay open, without closing and reopening.

Before updating

• No migrations.
• Restart the backend and frontend: a few seconds without the panel.

What changes

• New dashboard, with statistics.
  • Three groups of numbers: ONUs in the system, on the OLT and unauthorized; onboardings today, this month and in total (with imported ones counted separately); and ONUs online, not seen and with low signal. Each number opens the list of those ONUs.
  • Below: onboardings per month, ONU status, signal distribution, worst signals, traffic over 24 h, 7 or 30 days, health of each OLT, sortable PON ports and alerts by type.
  • Everything can be viewed for one OLT or for all of them.
• Powered-off ONUs no longer show as online. On a ZTE, the real state of each ONU is read (online, powered off, no light, power outage). Those that aren't working are marked down after 30 minutes, with their alert, and no longer sit at 0.00 dBm.
• Optical power no longer loses an entire port. With a slow OLT, such as the ZTE C320, a command's response could come back incomplete.
  • This left a port at 0 dBm, and “View configuration” or the unauthorized ONU scan came back empty.
  • Now the system waits for the device to finish responding.
• CPU, memory and temperature for ZTE OLTs, every 10 minutes, with their alarms.
• Fewer false “OLT down” alarms. The monitor tests SSH access every 5 minutes instead of every minute, and a single failed attempt while the OLT is still responding on the network no longer marks it as down. It also stops filling the device's alarm log with login entries.
• The optical reading waits its turn. If the OLT is busy with another operation, it waits a few seconds instead of skipping the cycle, and if it still skips it, it says so in the log.
• The same ONU registered twice on the OLT (an old registration on another port) no longer overwrites its own measurement. The backend log says which ones they are, so you can delete the old one from the device.
• The manufacturer on the ONU detail page: it comes from ONU Models, the catalog or the serial number, and says where it came from.
• Factory reset is added to the ZTE command suggestions.
• Reconciliation: an ONU imported with its name instead of its serial number only corrects itself if that name is unique on the OLT.

Before updating

• No migrations.
• Restart backend and frontend: a few seconds without the panel.
• On a ZTE, ONUs that are actually powered off are marked down after 30 minutes. Their alerts arrive together, just once.
• A ZTE's CPU, memory and temperature appear after 10 minutes.

What changes

• Import all of an OLT's ONUs at once (ONU › Reconciliation): you choose the plan once and every ONU the OLT reports is added, with its card, port, number and model. Nothing is sent to the OLT: they're already working there. The ones that can't be added say why.
• Each ONU's name on the OLT: on a ZTE, the name each ONU has on the device is read from the latest configuration backup. It shows in Reconciliation, is saved on import, appears in ONU › Authorized, and search finds it.
• An ONU imported with its name as the serial fixes itself: if an ONU was imported with its name instead of its serial (due to the 1.10.1 reading), when the real one arrives it gets its serial, location and model, keeping the plan and the subscriber.

Before updating

• Includes migration 062: the agent takes a backup and applies it.
• Restart the backend and frontend: a few seconds without the panel.
• The real ONUs appear in Reconciliation on the first optical reading, about 10 and a half minutes after restarting. Only then is it a good time to import.

What changes

• OLTs with slow SSH (ZTE C320 V2.1.0): no longer marked down every 10 minutes. The availability monitor and the optical power reading opened their SSH sessions in the same second; the device didn't finish both and the monitor marked it down for a minute. Now the optical reading is staggered, and the monitor doesn't open another session if the OLT already has one in progress.
• The per-port optical reading (ZTE) has more time to connect and to read all the ports. With this, ZTE ONUs appear in ONU › Reconciliation and the false findings close on their own.
• If an OLT's optical power reading fails, it's now recorded in the backend log.

Before updating

• No migrations.
• Restart backend and frontend: a few seconds without the panel.
• The first optical reading runs about 10 and a half minutes after restarting.

What changes

• ZTE OLTs: Reconciliation with the real ONUs. In 1.10.1, a C320 with firmware V2.1.0 filled Reconciliation with ONU names instead of serial numbers (and power levels of 0.01). Now the system reads first with the commands confirmed on that device, and over SNMP it only accepts what looks like a GPON serial number.
• False findings close on their own at the first good measurement, with the note “The OLT no longer reports it”. You don't need to ignore them one by one.
• From now on, on a ZTE, an ONU deleted from the OLT without having been imported also drops out of Reconciliation on its own.

Before updating

• No migrations.
• Restarts the backend and frontend: a few seconds without the panel.
• The ONUs show up, and the false findings close, at the next optical measurement (every 10 minutes).

What changes

• ZTE C300/C320 OLTs: ONUs are now visible. The system reads the ONUs and their optical power port by port, with the commands the C320 responds to (tested with the output of a real device running firmware V2.1.0). Before, none were read and ONU › Reconciliation stayed empty.
• Reconciliation is quicker to resolve: each ONU that is on the OLT but not in the system shows card, port, ID and model, and Import opens with that data already filled in.
• ONUs that the OLT reports with no signal (N/A) appear without a power value, not as zero.

Before updating

• No migrations.
• Restart backend and frontend: a few seconds without the panel.
• A ZTE's ONUs appear in Reconciliation on the next optical measurement (every 10 minutes).

What changes

• Billing integration (Integration › Billing): your billing system says who paid, and this one cuts off or reactivates the ONU, with the same profile change you make from the panel.
  • WispHub: suspending or activating a service there cuts off or reactivates the ONU instantly. With its API, a review also compares each service with its ONU.
  • Mikrowisp: it doesn't send notifications, so it's checked against its API periodically. You can run the review manually and it shows its progress.
  • eCloud-ISP: its “ONU at OLT level” cutoff notifies this system. Connecting it gives you the address, the endpoint and the secret to paste into eCloud-ISP.
  • An ONU suspended from the manager is never reactivated on its own, a canceled service is not cut off on its own, and the automatic review stops if it would have to suspend more than 10 ONUs (or 10 %) at once.
• FTP server in System settings: turn it on, turn it off and configure it without redeploying, with “Test connection”.
• Brand-new OLT out of the box: when adding an OLT, step-by-step initial setup guides for V-SOL, Huawei, ZTE, C-Data, FiberHome and TP-Link, with the commands filled in with your data.
• Panel footer with the system name and the version it's running.
• Fix: OLTs with old firmware (for example, ZTE C320 V2.1.0) can now be added over SSH. Previously the check failed with “Handshake failed: no matching C->S cipher”.
• Fix: the Tools menu in the OLT list closed with an error when opened (since 1.8.0).
• Fix: in Settings › FTP server, the toggle is no longer lost when you edit a field.

Before updating

• Includes migrations 060 and 061: the agent takes a backup and applies them. They don't change anything that already works; integrations start turned off.
• Restart the backend and frontend: a few seconds without the panel.
• For eCloud-ISP, its side needs migration V240 (still on a branch, unreleased).

What changes

• FTP users (Network settings): one account per OLT so the device can upload its own backup. The account can only upload, and only from that OLT's IPs. What it uploads shows up among the OLT's backups as “Uploaded by the OLT”, and binaries download with their extension.
• Automatic loop alarm: the system checks on its own whether an OLT reports a loop, opens the alarm, and closes it when the loop goes away. It ships turned off: turn it on in Thresholds › Loop detection.
• C-Data profile for FD12xx, FD13xx and FD16xx OLTs: commands from the official manual and SNMP recognition. C-Data OLTs added as “generic” can be moved to this profile by editing them.
• Shared CPEs in GenieACS: if a CPE that another operator had without a token reports your token, it becomes yours when you sync.
• Maps with Google: the official Google Maps logo in the attribution.
• Fix: an OLT whose configuration doesn't change no longer gets stuck with the “backup overdue” alert even though it backs up fine.

Before updating

• Includes migrations 057 to 059: the agent takes a backup and applies them.
• Restarts the backend and frontend: a few seconds without the panel.
• For OLTs to upload their backups, the FTP server has to be turned on in the stack (DEPLOY §11.54). Without it, the FTP users screen warns that it's off.

What changes

• ONU models with a catalog: “Add from catalog” adds 130 models from V-SOL, Huawei, ZTE, C-Data, FiberHome and TP-Link, with the manufacturer's official photo. It doesn't overwrite the models or photos you already have.
• OLT models (Network settings): the detail page of each OLT you use (PON ports, uplinks, management, firmware, power, photo), with a catalog of 79 OLTs from the same six manufacturers.
• Maintenance and logs, a new group under Monitoring:
  • Operation logs: now with the affected device, search by device, and export to CSV or TXT.
  • System logs: what happened on each OLT (outages, reboots, switchover to the failover IP, ports going down or up) and the OLT's internal log.
  • Alarms: OLT memory and voltage, ONU temperature, voltage and TX, and a notice of how many days are left before a declining signal crosses the threshold.
  • Diagnostic tools: find a device by serial number or MAC; ping, traceroute and loops from the OLT; remote access (CMS/MQTT). Plus Factory reset on each ONU's detail page.
  • Firmware and Maintenance windows moved to this group, at the same address.
• Fixes: the page no longer scrolls sideways with wide tables on narrow screens, and an uptime that arrived empty no longer opens a false reboot alert.

Before updating

• Includes migrations 052 to 056: the agent takes a backup and applies them.
• Restarts the backend and frontend: a few seconds without the panel.
• After installing:
  • Sync the CPEs (ACS · CPEs › Sync) so the MAC search can find them.
  • OLT memory and voltage, loops, remote access and factory reset require you to load each OLT's OIDs or commands (Edit OLT). Ping and traceroute come ready for Huawei and ZTE.

What changes

• DBA profiles and Line profiles, in Network settings:
  • Read from OLT shows the profiles the device has. The admin or support can request it.
  • New profile (admin only): first you see the exact commands that will be sent, and only when you confirm are they sent to the OLT. Then it reads again to check that the profile was created.
  • Works with Huawei (DBA and line) and ZTE C300/C320 (DBA). On vSOL you can load your device's query command in the OLT's advanced settings.
• Monitoring: the server block shows even when there are no OLTs yet, and “OLTs online” no longer says “All responding” when there are none.
• The panel's Spanish text now addresses you informally (“tú”) everywhere (“Carga”, “Puedes”, “Elige”).

Before updating

• Includes migration 051: the agent takes a backup and applies it.
• Restarts the backend and frontend: a few seconds without the panel.
• The first time, on each OLT: *Network settings › DBA profiles › Read from OLT*.

What changes

• Network configuration, a new submenu with:
  • Service profiles.
  • Zones, now on their own screen.
  • NAP boxes with comments, a photo of the box and “Locate on the map” (on your phone, with GPS).
  • ONU models, with import from the previous system's Excel file.
  • Firmware.
  • Logs: who did what and when, for the administrator only.
• OLT connection type when adding or editing it:
  • Public IP, or VPN through the MikroTik's WireGuard tunnel.
  • Optionally, a failover IP: if the primary one doesn't respond, the system connects through the other one and switches back on its own.
  • A VPN OLT whose IP isn't in the tunnel's networks is not saved.
• Monitoring, a new screen with:
  • OLT usage and health;
  • recent alarms;
  • traffic per uplink, PON port and ONU;
  • activity over the last 24 hours.
• Fixes a bug: editing an OLT could fail with “Internal server error”.

Before updating

• Includes migrations 048, 049 and 050: the agent takes a backup and applies them. Existing OLTs are set to “Public IP”, with no failover.
• Restart the backend and frontend: a few seconds without the panel.

What changes

• Connect Google Drive accepts the line exactly as rclone prints it (it starts with eyJ). Version 1.5.0 only accepted the token JSON and rejected what rclone authorize drive eyJzY29wZSI6ImRyaXZlLmZpbGUifQ returns.

Before updating

• Nothing. No migrations.
• Restart backend and frontend: a few seconds without the panel.

What changes

• System settings › Backups and restore:
  • You can see which backups exist and when the last full one finished.
  • You can back up on the spot.
  • You connect Google Drive by pasting the rclone token.
  • You can restore the entire system from a backup. First, a backup of the current state is taken, the previous database is kept, and if anything fails it rolls back on its own.
• Daily backups carry a label per run, so the panel can group their four parts.

Before updating

• Includes migration 047: the agent takes a backup and applies it.
• Restarts the backend and frontend: a few seconds without the panel.
• To use Backups you also need to rebuild the agent (1.2.0) and the backup image from Portainer (DEPLOY §11.42). Without that, the screen tells you so and everything else works the same.

What changes

• System settings (eCloud menu, platform staff only):
  • Mail server: configured from the panel, with a test send. Until it's configured, the usual environment variables are still used.
  • Google Maps API: Google Streets, Satellite and Hybrid as the base layer on maps. The key is tested against Google before it's saved.
  • Updates moves inside this menu.
• My account: by tapping your name at the bottom of the menu, you change your name, your photo and your password. Changing your password signs out your sessions on other devices.
• Fixed: changing the password with the current one mistyped could leave the request hanging, and each mistyped password counted twice against the attempt limit (also when enabling two-factor authentication or a passkey).

Before updating

• Includes three migrations (044, 045 and 046): the agent takes a backup and applies them.
• Restart backend and frontend: a few seconds without the panel. OLTs and subscribers are not affected.

What changes

• The panel installs on your phone like an app: an icon on the home screen, opens full screen and without the browser bar.
  • Android (Chrome): ⋮ menu › “Install app”.
  • iPhone / iPad (Safari): Share › “Add to Home Screen”. The panel shows a notice with these steps.
• No offline mode, on purpose: without a network, the app says there's no connection instead of showing stale network data.

Before updating

• Nothing. No migrations.
• Restart backend and frontend: a few seconds without the panel. OLTs and subscribers are not affected.

What changes

A minor diagnostic change, and the first version installed from the panel: until now, 1.3.1 was adopted by hand from the server.

• A network failure against GitHub only said “fetch failed”. Now it says which host and what the cause was (EAI_AGAIN, ECONNRESET…), which is what tells a passing outage apart from a DNS problem.

Before updating

• Nothing. No migrations.
• Restarts the backend and frontend: a few seconds without the panel. OLTs and subscribers are not touched.

What changes

Fixes the failure that prevented installing 1.3.0: in a private repository, the version's signed file couldn't be read.

• The release's version.json attachment is now downloaded through the GitHub API and with the token. The direct download URL returns 404 in a private repository, and the token is deliberately not sent to that host.
• It affected both the agent (installing) and the backend (the panel's “Check for versions” button).

Everything else is the same as 1.3.0: TR069, FTTx Maps, OLT onboarding and rolling back from the panel.

Before updating

• Nothing. The migrations are the same as in 1.3.0 (043 only adds columns).

What changes

This is the first published and signed version. From now on, the system updates from eCloud › Updates.

• TR069: WireGuard VPN to each ISP's private networks (VPN Users, LAN Networks), ACS tokens so each CPE stays with its operator, and the Servers screen (README §101).
• FTTx maps: outside plant design, ONU layers, coverage and network status, import from Excel (README §100).
• OLT onboarding: software version, PON type, Telnet and write community; OLT tools from the list (README §98 and §99).
• Updates: you can roll back to the previous version from the panel if the new one doesn't work out (README §102).

Before updating

• This version isn't “applied” from the panel: the agent adopts it once (DEPLOY §11.37), because the server currently runs a local build.
• New migrations: 043 (only adds columns).