Space map¶

The Map tab on a prefix's detail page lays out the address space inside that prefix as a grid of colored chips, so you can see at a glance what's free, what's used, and where you can carve out your next subnet.

What it shows¶

For each subnet size that fits inside the prefix, the map draws one chip per aligned subnet at that size:

• Green - free. No child prefix touches that block.
• Green with rose bars, rose outline - partly used. No child prefix covers the whole block, but smaller child prefixes sit inside it. Each rose bar runs the chip's full height and marks where one sits and how much of the block it takes, so a /26 holding one /28 shows a bar across its first quarter; the outline is drawn over the bars, so it stays whole. The block can't be allocated whole, but it still has free space one level down.
• Rose - used. The block is an existing child prefix, or sits inside one.
• Green with a small number - free, but already holding stray IPs. See Stray IPs below.
• Amber strip along the bottom - an IP range sits there. See IP ranges below.

A rose bar is the same fill as a used block, so a bar reads exactly as used as a rose chip. The map never paints free space as taken: children at most a pixel apart are drawn as one stretch, hatched when there are gaps in it, and when a block holds many small children scattered across it (more than 16 separate runs), each sixteenth of the block that holds any gets one thin bar, as wide as what it holds and placed where it sits. A /9 with a few dozen /24s spread through it shows a row of slivers over a green block, not a half-rose one.

Each row is one subnet size, labelled with a free-count (for example 5/8 free /19 subnets · 2 partly used · 1 holds an IP range), so you can read availability top to bottom.

2/2     free  /25 subnets    [ 10.20.30.0/25 ] [ 10.20.30.128/25 ]
4/4     free  /26 subnets    [ /26 ][ /26 ][ /26 ][ /26 ]
8/8     free  /27 subnets    8 cells
…

Hover any chip for what it is: Free, Free · 2 IPs inside, Free · range 10.196.196.10–50, 25% used · 10.196.238.128/28 (the child prefixes inside a partly used block), Existing prefix, or In 10.196.200.0/24 (the prefix a used block sits in).

Clicking a chip¶

A chip with one thing to do does it on click; a chip with several opens a menu.

A used block always offers Zoom into as well as Open, even when it is a single prefix: zooming maps that prefix's own space, so you can carve it up without leaving the map you started on. Blocks too small to split further (IPv4 /31, IPv6 /128) only offer Open.

New child prefix here opens the new-prefix form with that block pre-filled and the site and VRF inherited from the smallest prefix containing it, so you only confirm and save. Saving (or cancelling) brings you back to the map, zoomed where you were, with the new prefix drawn - so the next block is one click away. Register an IP here does the same.

The two create actions only show when you may add prefixes or IPs. Without either, a free block has just Zoom into …, so a click zooms straight in.

Zooming¶

Zooming re-roots the map at a block. A breadcrumb above the grid shows the path from the prefix down to the block on screen; click any step to go back to it, or the zoom-out button to go up one level.

The zoom path is part of the page address (?tab=map&zoom=10.196.224.0/19,10.196.238.128/26), and every zoom is a step in the browser history. The browser's Back button zooms out again; Back from a prefix you opened off the map, or from the create form, lands on the same zoomed view; and a reload or a shared link opens it directly.

When the block on screen is, or sits inside, an existing child prefix, the breadcrumb links to it (in 10.196.200.0/24). Free blocks there belong to that prefix: Register an IP here parents the new IP under it, not under the prefix the map started from.

Zooming is also how you get past the eight-bit limit below. A /18 draws its /19-/26 rows; zoom into a /26 and the map carries on to /31.

Zooming out of a prefix¶

A prefix inside a larger one (its masters, shown under its name) can also zoom out on its own Map tab, to see what sits around it. The breadcrumb then starts at the outermost master (10.196.192.0/18 › 10.196.238.0/24 › 10.196.238.128/28):

• Zoom out steps one bit at a time - the /28 to its /27, then the /26 - drawing its neighbours around it.
• The ▾ beside it lists every size from the outermost master down to the prefix, to jump several bits at once; this prefix goes back.
• A master on the breadcrumb zooms straight out to it.

The prefix is outlined wherever the map draws it, and each block holding it gets a dashed outline to follow down; picking the prefix's own chip (Back to …) returns to it. The view is part of the page address (?out=…), like the zoom path.

Everything on the map works from the keyboard: Tab to a chip, Enter to click it (or open its menu), arrow keys inside the menu. After a zoom, focus stays in the map.

Grid or aligned¶

Grid draws up to eight chips a line, each labelled. Aligned draws each row on one line, so every chip sits under the block it splits - a /22 under its own /20 - like slices of the row above; a chip too narrow for its whole address shows only its colour, and its tooltip names it. The choice is kept per user (space_map_layout).

Stray IPs¶

Sometimes a free block has no child prefix but does already contain individual IP addresses - IPs that are not inside any child prefix drawn on the map. That includes IPs parented straight to the prefix you are viewing: zoomed into 10.196.238.128/28, its own .129 and .130 count as stray, since no smaller prefix holds them. The space map marks those chips with a small count of the IPs inside, so loose addresses don't hide behind a plain "looks free" chip.

Create a prefix over one of those chips and the form warns that the existing IPs will be adopted - re-parented under the new prefix when you save, so the new prefix correctly owns them.

IP ranges¶

An IP range (a DHCP pool, a reservation) is not a prefix, so it doesn't make a block used: carving a subnet around a pool is normal. But the map shows where ranges sit, so you don't carve over one by accident. A free or partly used chip a range reaches gets an amber strip along its bottom edge, placed where the range is, and its tooltip names it (Free · range 10.196.196.10–50, or 3 ranges). The row label counts those chips.

A range that sits wholly inside a child prefix belongs to that prefix, and is not marked. Only ranges in the prefix's VRF count.

IPv6¶

The map works for IPv6 too, the same way. Because a v6 block has astronomically many subnets, the map shows a couple of nibble-aligned levels (a /64 shows its /68s and /72s) rather than thousands of cells. Partly used blocks zoom in on click like IPv4 ones, which is the natural way to navigate a sparse v6 plan one level at a time. A bar for a tiny child in a huge block (a /128 in a /56) still shows as a thin sliver.

How deep it draws¶

Each view draws at most eight bits below the block on screen (256 chips per row), stopping at /31 for IPv4 and /128 for IPv6; zooming carries on from there, down to /31 or /128.

Past those eight bits, Show /25 (512) under the last row adds the next row, one per click, down to your depth preference (a /16 goes all the way to its 32,768 /31s; a row stops being offered past 65,536 subnets). Such a row is drawn as one strip: used stretches in rose, partly used ones hatched, stray IPs a lighter green and IP ranges an amber edge, where they sit. Point at it - or Tab to it and use the arrow keys (Shift for bigger steps) - and its heading names the subnet there and its state; a click or Enter zooms in so that subnet becomes a chip of its own. The API is ?deeper=N on the space map, which also answers more with the row it would add next.

If that's more than you want to scan, set a shallower cap under Preferences → Space map (IPv4 / IPv6) - e.g. stop IPv4 at /24. The preference only ever makes a view shallower, never deeper than the eight-bit limit. It trims the views above it: with /24, a /18 shows /19-/24, and a zoom into a /20 shows /21-/24. Once you zoom into a block at or below the cap (a /24 here), there is nothing left for the cap to trim, so that view draws its full eight bits (/25-/31) rather than one row per click. IPv6 zooms keep their nibble steps the same way.

Limits¶

• Eight bits of chips per view. A very large block (say a /8) won't try to draw millions of /24s - it shows the next handful of sizes (IPv6 steps a nibble at a time, capped at 256 cells per row). Deeper rows come as strips, one click each, up to 65,536 subnets a row; zoom in for the rest.
• Down to the host boundary. IPv4 stops at /31; IPv6 at /128. A single IPv4 host (/32) isn't shown - that's what the IPs tab is for.
• Only child prefixes you can view count. A prefix hidden from you by site scope leaves its block looking free.
• IP ranges and IPs don't make a block used; they are marked on free and partly used chips (the amber strip, the stray-IP count).