Skip to content

v0.6.1

Choose a tag to compare

@anthomaxcool anthomaxcool released this 14 Nov 04:52

One YAASM driving several machines: servers on other PCs are listed, started,
backed up and moderated from one place, a cluster can span machines, and any
machine can take over as controller. Also: importing servers another tool set
up, yaasm uninstall, and Windows network shares for clusters run as a
service. Nothing changes on update: a machine only becomes a node once it is
given a node port and paired.

  • YAASM signs in to network shares itself (Windows). Run as a service,
    YAASM reaches a share as the computer, which a share on another PC turns
    down, so a cluster folder linked to one (mklink /D) stopped working the
    moment YAASM became a service. OPS › PATHS has a NETWORK SHARES card that
    lists the shares YAASM's folders lead to: give each the user and password
    of an account on the computer sharing it (or yaasm settings shares set),
    and YAASM signs in at every start, for itself and the game servers it
    starts. The password is kept encrypted and never shown again. yaasm doctor says when a saved share can't be opened.

  • A cluster folder YAASM can't open says why. A start used to fail with
    mkdir …\clusters: Cannot create a file when that file already exists
    when the cluster folder was a link to a network share YAASM's Windows
    service can't reach. It now names the link and the share, and says what
    to do. yaasm doctor checks each cluster's folder, and no longer passes a
    folder it can't open as one still to be created. The README says what a
    share on another PC means for a Windows machine running the service.

  • The top bar picks a machine. Once there is a node, the top bar's
    first crumb is a MACHINE picker, ALL MACHINES by default, beside the
    cluster one: FLEET, its heading and its START, RESTART and STOP ALL then
    cover that machine's servers. The server picker groups servers by machine
    as well as by cluster, so two machines' unclustered servers no longer
    read as one group.

  • One column on every page. A server's OVERVIEW and SETTINGS, AUTOMATION
    and its event and dynamic-config editors no longer split into two columns
    on a wide screen; they read top to bottom like the other pages.

  • OPS › SCREENS lists what can be hidden by page. FLEET, SERVER,
    CONSOLE, AUTOMATION and OPS each have their own group, instead of one long
    SECTIONS list mixing every page's parts.

  • Import servers another tool set up. A server from another server
    manager, or one copied over by hand, no longer has to be renamed and given
    a hand-written asm-config.json. yaasm server import "<folder>" lists the
    servers in a folder (or YAASM's own servers folder, with no folder) and
    imports one, or all with --all; IMPORT on the new-server page and the
    assistant's find_imports and import_server do the same. YAASM makes a
    new server on its own build, copies the old one's saves, config and logs
    in, and keeps its map, ports, session name and admin password. An id is
    made from the folder's name, since an id can't hold spaces. The old folder
    is left as it was, for you to delete once the new server runs. Mods are
    usually not written in the old server's files: give them when importing
    if the world uses any. yaasm doctor, which the installers run, now says
    when YAASM's servers folder holds servers it doesn't manage yet.

  • OPS folds. Each section's heading folds it away, and this browser
    remembers which; a link or search result to a section opens it again.
    The list under OPS jumps to each section, and the audit log, the longest,
    is now last.

  • A Windows node's port is checked against the firewall. yaasm doctor warns when Windows Firewall blocks the node port a controller
    reaches the machine on, and yaasm firewall lets it in, with the same
    rule as YAASM's web port. Before, a node whose web app stayed on this
    machine could be unreachable from its controller with nothing saying why.

  • A node can take over as controller. The controller keeps a copy of
    its accounts, tokens, player history, bans, events and node list on every
    node. If it's gone for good, yaasm node promote (or PROMOTE on a node's
    OPS › NODES) makes that node the controller from the copy, and the other
    nodes answer it at once. An old controller that comes back can no longer
    command them, and says it was replaced.

  • Clusters across machines. FLEET says, for each cluster with servers on
    more than one machine, whether they all see one cluster folder (ONE
    FOLDER) or not (FOLDER NOT SHARED, naming the machine that differs);
    yaasm cluster list says the same. When they don't, the OVERVIEW of each
    of its servers and yaasm doctor on the controller warn too: players
    moving between those machines lose their characters and items. The
    servers still start. The README's new Several machines
    section covers pairing and how to put a cluster's folder on a share.

  • Events reach every machine. An event made on the controller can name
    servers and clusters on any node; each node changes its own servers at
    the event's start and puts them back at its end, and AUTOMATION shows the
    run with every machine's servers. START NOW and END NOW act on every
    machine. A node's own events join the controller's at pairing; one that
    would clash with an event already there joins switched off, saying why.

  • One backup schedule, one set of update settings and one dynamic-config
    library for every machine.
    The controller's are sent to each node when
    saved, when a node is paired, and when a node comes back after being
    down; a node's own pages then refuse changes to them, pointing at the
    controller. A node's dynamic configs join the controller's library at
    pairing, under a name of their own when one is taken, and its servers
    keep exactly the settings they had. A node that couldn't take something
    says so on OPS › NODES.

  • The new-server form suggests island, not island-pve. A new server
    isn't PvE unless its settings make it so, and the suggestion read as
    though it were. yaasm server create's examples changed the same way.

  • Each machine's own settings, from the controller. OPS gains a
    MACHINE picker once there is a node: paths, YAASM's own update, plugins
    and the daemon panel below it then show and change that node's. A node
    server's pages use its own machine's builds, plugins and firewall check.
    On the command line, doctor, self-update, settings paths,
    version, build and plugin take --node. A node on an older version can still be
    updated from the controller.

  • yaasm service restart. Restarts YAASM's service the same way on
    Windows and Linux, leaving the game servers running. It first asks YAASM
    what it's in the middle of, and refuses while a backup, a download or a
    server starting or stopping would be cut short (--force goes ahead). yaasm doctor now says how to restart YAASM where it asks you
    to, and, when SteamCMD is missing under the service, to run
    yaasm service install again: the service's PATH is set at install, so
    a restart alone never found a newly installed SteamCMD.

  • Players and bans cover every machine. The controller's PLAYERS shows
    who is on its nodes' servers too, keeps one history for the whole fleet,
    and a ban, note or whitelist entry made there reaches every server on
    every node. When a node is paired, its own history and bans are merged
    into the controller's; from then on the node leaves them to its
    controller, and its PLAYERS refuses changes to bans and notes, pointing
    there. Unpairing gives it back what it had.

  • Servers on nodes. Once a machine is paired as a node, its servers
    show on the controller's FLEET beside its own, each marked with its
    machine, and everything on a server's page works on them as on the
    controller's: start, stop, the live chart, the console, settings, INI,
    backups. + NEW SERVER asks which machine once there is a node, as do
    yaasm server create --node and create_server. A server's id now has to
    be unique across every machine, and a node bringing an id the fleet
    already has is turned away at pairing with the list to move. A node that
    stops answering keeps its servers on FLEET, marked unreachable. Players,
    bans and the OPS panels are still each machine's own; they follow in
    later releases. A controller and its nodes must all be updated to this
    release together.

  • yaasm uninstall. Removes what the installer put on the machine: the
    service, the binary and its folder, its place on PATH, the firewall rules
    yaasm firewall made and, on Linux, the downloaded GE-Proton. It lists
    everything first and asks. YAASM's data stays; --purge deletes it too,
    with the yaasm account on Linux, for a machine done with YAASM or a
    fresh install from scratch. Game servers have to be stopped first.

  • Deleting a server works on Windows while another server on the same
    build is running.
    It used to fail with Access is denied on one of the
    game's DLLs, after deleting part of the server's folder; delete it again
    to finish. Removing old builds and old plugin releases could fail the
    same way and leave their files behind.

  • The web app fits an iPad in landscape and a phone turned sideways.
    Between about 900 and 1100 pixels wide, pages ran off the right edge of
    the screen: the fleet's tiles and START/RESTART/STOP ALL, a server's
    SETTINGS switches, the event editor's dates. In that range the search
    field is now an icon, as on a phone, and the event editor is one column.
    On an iPhone in landscape, text no longer grows and the sidebar keeps
    clear of the notch.

  • An event's start date sticks. A repeating event set to begin on a later
    date used to start at the schedule's next match instead, and the editor
    showed that date after saving, with no way to put the chosen one back. The
    start picked in the dates view is now kept, and no run begins before it
    (from in yaasm schedule events JSON). Moving the start no longer moves
    the end, unless the start lands on or after it.

  • EDIT beside a server's name, at the top of its page, changes the name
    players see in the server browser. It's written into the game's config, as
    the INI tab would; a running server shows the new name from its next start,
    and the page says so.

  • A server's id can be changed, from MOVE SERVER on a stopped server's
    SETTINGS, yaasm server move, or move_server for assistants. The id
    is its folder under servers/ and what YAASM knows it by; the new-server
    form now calls it ID rather than NAME, beside the session name players
    see. Its folder moves, and its backups, schedules, plugins, the bans YAASM
    applied there, its players' history and the events that list it follow.
    On Windows, run yaasm firewall again afterwards.

  • Nodes, the first step towards one YAASM driving several machines. A
    machine can pair with another YAASM, its controller: set
    YAASM_NODE_LISTEN_ADDR on it, press PAIR on its OPS › NODES (or run
    yaasm node pair), and type the code it shows on the controller's
    OPS › NODES › ADD (or yaasm node add). The controller lists each node
    with whether it's up and its version. The link is always encrypted, and
    the code proves each machine to the other. Servers on nodes come in a
    later release.

Fleet

  • CPU and RAM on every fleet card. Each running server shows what it uses
    (CPU 1.5% RAM 7.8 GB), and the header shows the machine's memory. The
    line stays grey while things are normal and turns amber or red when a server
    is using most of the machine's CPU or the machine is running out of memory,
    so a healthy fleet has nothing to read. It can be hidden from OPS › SCREENS.

  • A server's RAM chart leads with the amount. The top of the chart now
    reads 7.8 GB / 64, as on the fleet card; the share of the machine moved to
    the legend, next to CPU's.

  • Building from source on Windows no longer gives a blank page. A build
    made after the sources changed produced a web app that never started.
    Releases and builds on Linux weren't affected.