flagon-io/g1t

public

Where people and agents ship software together. The open-source git platform for the whole job: issues, agents, checks and deploys to the edge.

git agents cloudflare rust forge mcp

Commit

Room for public repositories' ubuntu-latest jobs: 30 sandboxes on the 4-core machine, a pool of 60

Part of #169. Since #170, every public repository's workflow jobs run on the 4-core class, which had room for 10, where most of them had 20 on the standard class before. Deploy #436's plan job had already failed once with "no container instance that can be provided". - services/runner/wrangler.jsonc: Sandbox4Core max_instances 10 to 30. All three classes together hold 60 sandboxes: 180 vCPU, 600 GiB, 1,080 GB, inside the account's 1,500 vCPU, 6 TiB, 30 TB. - services/actions: HOSTED_POOL 40 to 60, so the per-workspace burst follows the larger pool. - The compute and Cloudflare docs say the new numbers.

Checked:: cargo test -p g1t-actions-service; the runner's and the deploy tool's tests; prettier.
G1t-Session:: ses_01m57b7r6jew3aqgyfmhw6yah6

syntaqxcommitted 3h agoParentf405562Browse files

Merged inRoom for public repositories' ubuntu-latest jobs: 30 sandboxes on the 4-core machine, a pool of 60 #171Why it was written

54	54
55	55		\| \| \|
56	56		\| --------------------------- \| ------------------------------------------------------------------------------------------------- \|
57		−	\| Hosted jobs running at once \| 4 per workspace always, up to 12 while the pool of 40 sandboxes keeps 8 free for other workspaces \|
	57	+	\| Hosted jobs running at once \| 4 per workspace always, up to 12 while the pool of 60 sandboxes keeps 8 free for other workspaces \|
58	58		\| A hosted job's timeout \| 60 minutes at most \|
59	59		\| A self-hosted job's timeout \| 24 hours at most \|
60	60		\| Logs \| 4 MiB a job, kept 90 days \|

342	342		16 GB), and so is `runs-on: g1t-2core`, in a class of its own; jobs that
343	343		ask for `runs-on: g1t-4core`, and `ubuntu-latest` jobs in a public
344	344		repository, get `standard-4` (4 vCPU, 12 GiB, 20 GB). At
345		−	the `max_instances` in `services/runner/wrangler.jsonc` (20, 10 and 10),
346		−	40 sandboxes at once hold 100 vCPU, 360 GiB of memory and 680 GB of disk,
	345	+	the `max_instances` in `services/runner/wrangler.jsonc` (20, 10 and 30),
	346	+	60 sandboxes at once hold 180 vCPU, 600 GiB of memory and 1,080 GB of disk,
347	347		which your account's Containers limits for live instances must allow.
348	348
349	349		Set the runner Worker's `DOCKER` variable to `off` to give workflow jobs no

65	65		/// The most one workspace may run at once while the pool has room to spare.
66	66		pub const BURST_PER_WORKSPACE: u32 = 12;
67	67		/// Sandboxes g1t's own machines have between them (services/runner's
68		−	/// wrangler.jsonc: 20 standard, 10 g1t-2core, 10 g1t-4core).
69		−	pub const HOSTED_POOL: u32 = 40;
	68	+	/// wrangler.jsonc: 20 standard, 10 g1t-2core, 30 g1t-4core, which every
	69	+	/// public repository's ubuntu-latest jobs run on).
	70	+	pub const HOSTED_POOL: u32 = 60;
70	71		/// Kept free while workspaces run past their share, so another workspace's
71	72		/// first jobs still start at once.
72	73		pub const POOL_HEADROOM: u32 = 8;

9	9		// types: standard-3 (2 vCPU, 8 GiB, 16 GB of disk) for the standard
10	10		// machine and g1t-2core, which is kept as a name of it so workflows that
11	11		// ask for it keep working, and standard-4 (4 vCPU, 12 GiB, 20 GB) for
12		−	// g1t-4core. At these max_instances, 40 sandboxes at once hold 100 vCPU,
13		−	// 360 GiB and 680 GB, well inside the account's limits for live
14		−	// instances (1,500 vCPU, 6 TiB of memory). crates/contracts
	12	+	// g1t-4core, which is also what ubuntu-latest gets in a public
	13	+	// repository. At these max_instances, 60 sandboxes at once hold 180 vCPU,
	14	+	// 600 GiB and 1,080 GB, well inside the account's limits for live
	15	+	// instances (1,500 vCPU, 6 TiB of memory, 30 TB of disk). crates/contracts
15	16		// (INSTANCE_TYPES) and services/billing (keeper.rs) say the same sizes.
16	17		// scripts/deploy.mjs deploys with each image set to the reference it
17	18		// built and pushed ("The runner's images" in
⋯
22	23		"class_name": "AttemptSandbox",
23	24		"image": "./Dockerfile",
24	25		// Workflow jobs with `runs-on: ubuntu-latest` (and every other
25		−	// hosted label), and all other work: 2 vCPU, 8 GiB, 16 GB.
	26	+	// hosted label) in a private repository, checks, the merge queue and
	27	+	// all other work: 2 vCPU, 8 GiB, 16 GB.
26	28		"instance_type": "standard-3",
27	29		"max_instances": 20,
28	30		// A deploy that changes the image or these settings replaces every
⋯
41	43		"rollout_active_grace_period": 7200
42	44		},
43	45		{
44		−	// Workflow jobs with `runs-on: g1t-4core`: 4 vCPU, 12 GiB, 20 GB.
	46	+	// Workflow jobs with `runs-on: g1t-4core`, and `ubuntu-latest` (and
	47	+	// every other hosted label) in a public repository: 4 vCPU, 12 GiB,
	48	+	// 20 GB. Every public repository's workflow jobs run here, so it has
	49	+	// the most room.
45	50		"class_name": "Sandbox4Core",
46	51		"image": "./Dockerfile",
47	52		"instance_type": "standard-4",
48		−	"max_instances": 10,
	53	+	"max_instances": 30,
49	54		"rollout_active_grace_period": 7200
50	55		}
51	56		],

flagon-io/g1t

Room for public repositories' ubuntu-latest jobs: 30 sandboxes on the 4-core machine, a pool of 60

Comments