Every change.

Public pages that search engines can read, a sign-up check against throwaway mail addresses, and clearer answers when a custom domain or its certificate cannot be set up. Most of the rest is staff-side work.

Changed

• The public pages are ready to carry their text. Pricing, the legal pages and the documentation can be rendered on the server, so search engines get their text already in the page. The renderer does not run until the hidencloud-ssr unit is installed on the host; until then the pages render in the browser as before. Search engines may now fetch the styles, fonts and images those pages use.
• Sign-ups from throwaway mail addresses are refused. If the email of your Discord account is at a disposable mail domain, the panel does not open an account for it (acceptable use policy 4.7).
• Signing in with no page to go back to takes you to the dashboard.
• If your Discord account does not take direct messages from the bot, the panel waits a day before it tries again. Your notices still arrive in the panel.
• Node names start with a capital letter wherever the panel shows them.
• In /admin, by default a ban marks the account and the devices only it has used; the IP, a shared fingerprint and the ASN are ticked by hand. A ticket nobody has answered within a day is sent to the support team.

Fixed

• A custom domain that does not point here says so. The panel checks where the domain points before it asks for a certificate, and tells you when it is behind another CDN, on another host or has no address.
• A refused certificate says why. Under a rate limit, the page says until when, and the retry waits for it.
• The subdomain page says a custom domain does not serve a website, and points to the reverse proxy for one.
• A banned account always sees its ban and the way to appeal it.
• A server suspended for abuse can no longer be renewed, and Renew all leaves it out.
• A server under an abuse suspension, or quarantined with an appeal pending, is no longer deleted at expiry before staff answer, and a quarantine lifted later gives the renewal time back.
• An upload that cannot reach the node stops and says so instead of waiting.
• A locked backup says it is locked in words, in your language.

Security

• security.txt is served with an expiry date counted from the day it is read, so it never runs out.

One renewal reminder a cycle instead of four, your renewals in your security log, and world and player files that can no longer keep the panel busy. Most of the rest is staff-side work on the abuse queue.

Changed

• One renewal reminder a cycle. You get it when your server's 24-hour renewal window opens. The reminder two days before, the last call on the due day and the missed-window notice are gone. The suspension and termination notices have not changed.
• A sanction keeps the text you were first told. If staff reclassify it later, your notice stays as it was, and your data export says when it was reclassified and what it said before.
• An appeal against a suspension a person issued shows the evidence. Staff can name the finding the suspension answers, when they issue it or later, and the appeal page then shows what was found, as it does for an automatic one.
• Deleting a server lifts its sanctions. They no longer stay live on a server that is gone, and they still count on the account's record.
• The worlds page lists up to 64 world folders. A world past them can be opened by typing its name.
• The world pages and the Jars page share a limit of 60 reads a minute for each account on each server.
• In /admin, staff can claim an abuse finding so others see who holds it, filter the abuse queue by type, severity and account, and see the servers of a banned owner counted as suspended everywhere.

Security

• Your renewals are in your security log. Each renewal you ask for shows on /account/security with its address, country, network and whether it came from a datacenter.
• World and player files are read in turns, and a file that unpacks past a size ceiling is refused, so one account can no longer take all of the panel's time with a crafted file.

Fixed

• The Player Manager and the worlds page say when a file cannot be read. They used to show default values, or an old copy of the player.
• A restore result the node sends late no longer ends a restore that is running now.
• Signing out of the staff sign-in ends that session without asking the person to pick one.

The panel fits a phone, staff can set upload and download ceilings for an egg and for one server, and the infected-plugin notice says which jar it found and what to do about it.

Four of the changes below need the node's daemon to be up to date, and each node gets it before this version goes live: the network ceilings, the jar's name in the infected-plugin notice, the exact room in a node's shared disk, and the notice that a node has no room to start your server.

Added

• Upload and download ceilings for an egg and for one server. Staff set an egg's ceiling in Mbit/s from its page in /admin, and can give one server its own, with a reason and an optional expiry up to a year out. The server's page in /admin shows the ceiling in force and where it comes from. A running server takes the change without a restart.
• The infected-plugin notice names the jar. It says "We found it in plugins/X.jar", or how many other jars, so you do not have to guess which of your plugins to remove.
• Report a problem from the phone menu. Inside a server, the side menu on a phone has "Report a problem", which opens the report form the console hides on a small screen.
• The language cards in your account settings show a flag beside the language name.
• Every edit of a node in /admin is recorded in the activity log, with each field it changed before and after.

Changed

• Node.js 24 is the default image. 22 and 20 stay, and 18, which reached end of life, is gone. A server on 18 keeps running and moves to 24 when you reinstall it, or when staff rebuild it. Its Startup page shows 18 as the current image, no longer listed, and the reinstall confirmation says which version the server comes back on.
• A rebuild from /admin says which image it moves a server to. When staff rebuild a server whose image is no longer offered, the confirmation names the image it is on and the one it moves to, and the owner gets a notice that the image changed.
• Go 1.26 is the Golang image. The egg offered Go 1.21, 1.22 and 1.23, all past end of support, and now runs Go 1.26 only. A server on one of the old versions keeps running and moves to 1.26 when you reinstall it, or when staff rebuild it.
• Regions are shown by their name, in your language. The flag beside a region is gone (Windows drew two letters instead of it), and the us region is now called North America (Norteamérica in Spanish). The create wizard, the databases page and the Onyx card all name the regions in the language you read the panel in.
• The infected-plugin notice says what to do. It gives the steps to clean the plugin and says whether your server is already restricted because of another finding. On a server whose network is isolated, it tells you to clean the plugin now, since your files and console stay open. On a server suspended for another reason, it says the files cannot be changed now and which jars to leave out if you download them.
• The create wizard says what each server type lets you choose. A Node.js, Python, C# or Go server reads about the runtime, an egg with fixed versions about the version, and only Minecraft Java about the world and the Jars page.
• Staff can set a node's disk overcommit up to 1000000%.
• Staff can put a node in Oceania from the node pages in /admin, and those pages name the region in your language too.
• A node that keeps its servers in one shared disk image takes only what that image can hold. It used to accept servers until the image filled up, and then every server on it stopped at once. A node whose daemon could not read that image takes no new servers until it can. Servers already on the node are not touched.
• A server created from /admin is checked against its node. Staff are told which check refused it (memory, seats, or the node's shared disk), and the create wizard in /admin marks those nodes as unavailable and says why. An admin can still place it on a node short of memory or seats, or whose daemon has gone quiet, by giving a reason, which goes to the activity log. A full shared disk cannot be overridden.
• A staff sign-in to /admin lasts one hour instead of twelve.

Fixed

• The panel and the public pages fit a phone. The console toolbar fits a 360px screen, the report button no longer covers the console, a profile picture that does not load shows your initial, and the copy icon beside a long server address keeps its size. The public header keeps one row down to a 320px screen, with the logo lined up with the page and a short sign-up button ("Únete" in Spanish) that is easy to tap. The status strip fits the screen. The links of the phone menu and the console icons are at least 44px.
• Every public page says when sign-ups are paused. Only the home and pricing pages did, and the others still offered the sign-up button.
• The Advanced tab of a server reaches the node again. It said the node could not be contacted, on every server.
• In a ticket, the copy button no longer covers the message's text on a touch screen.
• In a ticket, the Send button shows its spinner in the light theme while your reply is sending.
• The Discord logo has both eyes wherever the panel draws it.
• In your account settings, the name of the selected language reads in the light theme.
• The infected-plugin notice no longer names a jar that staff marked as a false positive.
• The notice about a paused Onyx reservation names the region in your language, not by its code.
• An expiry set in /admin for a server's network ceiling or for an egress entry is stored as the moment it names, whatever time zone it was sent from, and a date more than a year out is refused with a message instead of an error. The dialog to remove a server's ceiling reads as a sentence, and its events have labels in /admin/activity.
• An upload ceiling above the download one is refused, on an egg's page and on a server's override in /admin. The node already cut the upload down to the download, so the page showed a number that was not the one in force.
• The server's pages say when its node has no room to start it. A notice stays on every page of the server and Start and Restart are disabled, with the reason. The server used to sit on "Starting…" with nothing said.
• Staff see the report button, and a report they file is handled as theirs.
• A file renamed inline is renamed once. Pressing Enter sent the rename twice, and the second one showed an error for a rename that had worked.
• The SFTP importer says when the other host stops answering. An import whose host dropped used to end with a generic error, or as completed with files missing. It now fails and says the host stopped answering.
• A node posture incident keeps the list of required items that failed, after the node recovers.
• A banned account that signs in with Discord is turned away before its session opens, with the same notice /login shows, and the sign-in is recorded as refused instead of as a success. An account with two-factor authentication is no longer shown the code screen first.
• Allowing a pair of accounts in /admin undoes their prohibition. The multi-account ban and its warning are lifted when the staff member can manage bans; otherwise the confirmation says how many are still active and that they are lifted in /admin/bans. A server the prohibition put in the 48-hour download window comes back with a new renewal date if the window has not run out. Before, allowing the pair only stored the decision. Marking the accounts of a finding's group as different people does the same for the group, including the accounts the rule banned with it. A related account whose ban nothing ties to the group stays banned, and the confirmation says how many are left to review in /admin/bans.
• The ban form in /admin/bans offers the device the account was last seen with, as the user's page in /admin does. It offered the last device recorded for the first time, so an account that went back to an earlier browser was offered one it no longer used. A confirmed mining case names the same device the ban forms tick.

A small release. The panel can now check that your own domain points at the proxy of its region, and staff see the same room on each node that the panel uses to place servers.

Added

• Own domains are checked against the proxy of their region. Every ten minutes the panel looks up each verified own domain in public DNS and records whether it points at the proxy of its server's region: a CNAME through the region's proxy name, or A records that are all addresses of that proxy. An IPv6 record that leads somewhere else keeps the domain from counting. A DNS outage changes nothing; only a clear answer that points elsewhere does.

Fixed

• The admin server wizard shows each node's real room. Free memory and disk now take the node's overcommit into account, as placement does, and each node shows how many servers it takes ("3/12 servers") and says "No seats" when it is full. Staff can still pick a full node.

A short release. It stops one person from taking several free servers through several accounts, and most of the rest is staff-side work that does not show here.

Changed

• A second account does not get a second server. An account linked to another account that already has a server cannot create a second one. The create page says why and lets you ask a person to review the link; if the accounts turn out to belong to different people, you can create it.
• Several accounts of one person, each with a server: the oldest account keeps its server and gets a warning, and the others are banned.
• Phantom, the software that runs your server on its node, is called by its name in every message, instead of "Wings" or "the daemon".

Fixed

• Your security activity records the country of every sign-in again, including sign-ins from an address the panel had seen before.