lfsx_server/
config.rs

1use axum::http::{HeaderMap, header};
2
3use std::net::SocketAddr;
4use std::path::PathBuf;
5use std::time::Duration;
6
7use crate::model::Action;
8use crate::namespace::Namespace;
9
10#[derive(Debug, Clone)]
11pub struct Config {
12    pub bind: SocketAddr,
13    pub storage_root: PathBuf,
14    pub public_url: Option<String>,
15    pub action_lifetime: u32,
16    pub gc_grace: Duration,
17    pub staging_max_age: Duration,
18    // How long a lock may go untouched before anyone can take it. Unset means
19    // never, which is what happened before this existed and what a team that has
20    // not thought about it yet should keep getting.
21    pub lock_max_age: Option<Duration>,
22    pub max_object_size: Option<u64>,
23    pub repo_quota: Option<u64>,
24    pub compression: Option<i32>,
25    // A path rather than the key itself: a key in the environment is in the pod
26    // spec, in `docker inspect`, and in every log that dumps the environment. A
27    // file comes from a Kubernetes Secret mount or FerrVault without any of that.
28    pub encryption_key_file: Option<PathBuf>,
29    pub storage: Storage,
30    pub auth: Auth,
31}
32
33#[derive(Debug, Clone)]
34pub enum Storage {
35    Local,
36    // Endpoint, bucket and credentials all have to be there: a bucket the server
37    // cannot reach is a server that answers every push with an error, and
38    // discovering that on the first upload rather than at boot is the wrong
39    // order.
40    Bucket {
41        endpoint: String,
42        bucket: String,
43        region: String,
44        access_key: String,
45        secret_key: String,
46        path_style: bool,
47        // Whether a download is redirected to the bucket instead of streamed
48        // through this server. Off by default: the streamed path is the one
49        // that counts bytes, serves ranges and holds the ceiling, and an
50        // operator should choose to give those up rather than discover it.
51        presign: bool,
52        // Whether locks can be taken here. Not read from the environment: it
53        // starts true and the startup probes turn it off, the same way they turn
54        // `presign` off, when the store will not prove it can arbitrate between
55        // two writers racing for the same key.
56        locking: bool,
57    },
58}
59
60impl Storage {
61    fn from_env() -> Self {
62        if std::env::var("LFSX_STORAGE").as_deref() != Ok("s3") {
63            return Self::Local;
64        }
65
66        let required = |name: &str| {
67            std::env::var(name)
68                .ok()
69                .filter(|value| !value.is_empty())
70                .unwrap_or_else(|| panic!("LFSX_STORAGE=s3 needs {name}"))
71        };
72
73        Self::Bucket {
74            endpoint: required("LFSX_S3_ENDPOINT"),
75            bucket: required("LFSX_S3_BUCKET"),
76            region: std::env::var("LFSX_S3_REGION").unwrap_or_else(|_| "us-east-1".into()),
77            access_key: required("LFSX_S3_ACCESS_KEY"),
78            secret_key: required("LFSX_S3_SECRET_KEY"),
79            path_style: std::env::var("LFSX_S3_PATH_STYLE").as_deref() != Ok("false"),
80            presign: std::env::var("LFSX_S3_PRESIGN").as_deref() == Ok("true"),
81            locking: true,
82        }
83    }
84}
85
86#[derive(Debug, Clone)]
87pub enum Auth {
88    Forge {
89        provider: Provider,
90        api_url: String,
91        cache_ttl: Duration,
92        rejection_ttl: Duration,
93        // Lookups a minute this server will spend on the forge, counted only
94        // when neither cache could answer. None is no ceiling at all.
95        lookup_budget: Option<u32>,
96        // Whether a request with no credentials is resolved against the forge
97        // instead of refused. On by default, because that is what cloning a
98        // public repository does everywhere else.
99        anonymous_read: bool,
100    },
101    Disabled,
102}
103
104#[derive(Debug, Clone, Copy, PartialEq, Eq)]
105pub enum Provider {
106    Github,
107    Gitlab,
108    // Gitea and Forgejo, which are one API: Forgejo is a fork of Gitea and
109    // answers the same routes, so the only thing that tells two instances apart
110    // is the root they are reached at.
111    Gitea,
112}
113
114impl Provider {
115    // None where there is no such thing as the instance.
116    //
117    // github.com and gitlab.com are where a repository is unless the operator
118    // says otherwise, so defaulting there is nearly always right. Gitea is
119    // software rather than a place. gitea.com exists, but an operator who names
120    // this provider is almost certainly running their own, and quietly resolving
121    // their namespaces against a stranger's forge is worse than not starting: a
122    // public repository there that happens to share a name would hand out
123    // anonymous read on objects it has nothing to do with.
124    fn default_api_url(self) -> Option<&'static str> {
125        match self {
126            Self::Github => Some("https://api.github.com"),
127            Self::Gitlab => Some("https://gitlab.com/api/v4"),
128            Self::Gitea => None,
129        }
130    }
131
132    fn api_url_variable(self) -> &'static str {
133        match self {
134            Self::Github => "LFSX_GITHUB_API_URL",
135            Self::Gitlab => "LFSX_GITLAB_API_URL",
136            Self::Gitea => "LFSX_GITEA_API_URL",
137        }
138    }
139}
140
141const CACHE_TTL: Duration = Duration::from_secs(60);
142const REJECTION_TTL: Duration = Duration::from_secs(10);
143// Generous enough that a busy server never meets it, since a lookup is one
144// distinct token against one repository per cache lifetime rather than one per
145// request, and tight enough that a flood costs ten a second instead of whatever
146// the network will carry.
147const LOOKUP_BUDGET: u32 = 600;
148const GC_GRACE: Duration = Duration::from_secs(14 * 24 * 60 * 60);
149const STAGING_MAX_AGE: Duration = Duration::from_secs(24 * 60 * 60);
150
151impl Config {
152    pub fn from_env() -> Self {
153        let bind = std::env::var("LFSX_BIND")
154            .ok()
155            .and_then(|raw| raw.parse().ok())
156            .unwrap_or_else(|| SocketAddr::from(([0, 0, 0, 0], 8080)));
157
158        let storage_root = std::env::var("LFSX_STORAGE_ROOT")
159            .map(PathBuf::from)
160            .unwrap_or_else(|_| PathBuf::from("/var/lib/lfsx"));
161
162        let public_url = std::env::var("LFSX_PUBLIC_URL")
163            .ok()
164            .filter(|url| !url.is_empty())
165            .map(|url| url.trim_end_matches('/').to_owned());
166
167        Self {
168            bind,
169            storage_root,
170            public_url,
171            action_lifetime: 1800,
172            gc_grace: seconds("LFSX_GC_GRACE").unwrap_or(GC_GRACE),
173            staging_max_age: seconds("LFSX_STAGING_MAX_AGE").unwrap_or(STAGING_MAX_AGE),
174            lock_max_age: seconds("LFSX_LOCK_MAX_AGE"),
175            max_object_size: bytes("LFSX_MAX_OBJECT_SIZE"),
176            repo_quota: bytes("LFSX_REPO_QUOTA"),
177            compression: compression(),
178            encryption_key_file: std::env::var("LFSX_ENCRYPTION_KEY_FILE")
179                .ok()
180                .filter(|path| !path.is_empty())
181                .map(PathBuf::from),
182            storage: Storage::from_env(),
183            auth: Auth::from_env(),
184        }
185    }
186
187    pub fn base_url(&self, headers: &HeaderMap) -> String {
188        if let Some(configured) = &self.public_url {
189            return configured.clone();
190        }
191
192        let scheme = headers
193            .get("x-forwarded-proto")
194            .and_then(|value| value.to_str().ok())
195            .and_then(|value| value.split(',').next())
196            .map(str::trim)
197            .filter(|scheme| !scheme.is_empty())
198            .unwrap_or("http");
199
200        let authority = headers
201            .get(header::HOST)
202            .and_then(|value| value.to_str().ok())
203            .map(str::trim)
204            .filter(|host| !host.is_empty())
205            .unwrap_or("localhost");
206
207        format!("{scheme}://{authority}")
208    }
209
210    pub fn object_url(&self, base: &str, ns: &Namespace, oid: &str) -> String {
211        format!("{base}/{ns}/objects/{oid}")
212    }
213
214    pub fn verify_url(&self, base: &str, ns: &Namespace) -> String {
215        format!("{base}/{ns}/objects/verify")
216    }
217
218    pub fn action(&self, href: String) -> Action {
219        Action {
220            href,
221            header: None,
222            expires_in: self.action_lifetime,
223        }
224    }
225
226    pub fn signed_action(&self, href: String, headers: Vec<(String, String)>) -> Action {
227        Action {
228            href,
229            header: Some(headers.into_iter().collect()),
230            expires_in: self.action_lifetime,
231        }
232    }
233}
234
235// Opt in, not opt out. Serving objects to a caller with no credentials at all is
236// a decision an operator should make on purpose: it costs them the bandwidth of
237// anyone who finds the endpoint, on a server whose whole job is to move files
238// measured in gigabytes. Nothing confidential is at stake, since a request with
239// no credentials is still resolved against the forge and a private repository is
240// still refused, but "anyone may pull from you" is not a sensible thing to
241// inherit by default.
242//
243// Only the exact string opens it. A typo, an empty value or a `1` leaves it
244// closed, because the failure that matters here is the one that opens the door
245// when nobody meant to.
246fn anonymous_read(value: Option<&str>) -> bool {
247    value == Some("true")
248}
249
250impl Auth {
251    fn from_env() -> Self {
252        if std::env::var("LFSX_AUTH").as_deref() == Ok("disabled") {
253            tracing::warn!(
254                "LFSX_AUTH=disabled — every request is accepted, run this on a trusted network only"
255            );
256            return Self::Disabled;
257        }
258
259        let provider = provider(std::env::var("LFSX_AUTH").ok().as_deref());
260
261        Self::Forge {
262            provider,
263            api_url: api_url(
264                provider,
265                std::env::var(provider.api_url_variable()).ok().as_deref(),
266            ),
267            cache_ttl: seconds("LFSX_AUTH_CACHE_TTL").unwrap_or(CACHE_TTL),
268            rejection_ttl: seconds("LFSX_AUTH_REJECTION_TTL").unwrap_or(REJECTION_TTL),
269            lookup_budget: lookup_budget(std::env::var("LFSX_AUTH_LOOKUP_BUDGET").ok().as_deref()),
270            anonymous_read: anonymous_read(std::env::var("LFSX_ANONYMOUS_READ").ok().as_deref()),
271        }
272    }
273}
274
275// Zero is the one value that cannot mean what it says. A ceiling of no lookups
276// is a server that refuses every caller it has not already seen, so it is read as
277// the operator turning the ceiling off, which is the only other thing they could
278// have meant. Anything unparseable is the default rather than a refusal to start:
279// this bounds a cost, and getting it wrong should not take the server down.
280fn lookup_budget(value: Option<&str>) -> Option<u32> {
281    match value.map(str::trim).map(str::parse::<u32>) {
282        Some(Ok(0)) => None,
283        Some(Ok(budget)) => Some(budget),
284        Some(Err(_)) | None => Some(LOOKUP_BUDGET),
285    }
286}
287
288// Anything unrecognised is GitHub, which is what an operator who set nothing
289// almost certainly meant. Forgejo is named alongside Gitea because they are one
290// API, and somebody running Forgejo should not have to know it began as a fork.
291fn provider(value: Option<&str>) -> Provider {
292    match value {
293        Some("gitlab") => Provider::Gitlab,
294        Some("gitea") | Some("forgejo") => Provider::Gitea,
295        _ => Provider::Github,
296    }
297}
298
299// The trailing slash matters: every route is built by appending to this, so one
300// left on the end produces `//repos/...`, which some forges answer and others do
301// not, and the ones that do not answer 404 for a repository that is right there.
302fn api_url(provider: Provider, configured: Option<&str>) -> String {
303    let variable = provider.api_url_variable();
304
305    configured
306        .map(str::to_owned)
307        .or_else(|| provider.default_api_url().map(str::to_owned))
308        .unwrap_or_else(|| {
309            panic!(
310                "{variable} must be set: a self-hosted forge has no default API root, and guessing \
311                 one would resolve your repositories against somebody else's"
312            )
313        })
314        .trim_end_matches('/')
315        .to_owned()
316}
317
318// Unset means unlimited, which is what a server on its own volume wants. Zero
319// would refuse every push, so it is read as a typo rather than as a policy
320// nobody would choose deliberately.
321// zstd level 3 is the default because it is the one that costs nothing you can
322// measure: it compresses faster than a spinning disk writes, and the meshes and
323// uncompressed raster that make up most of an LFS store give most of their
324// ground at any level. Higher levels are there for a store that is short on
325// room rather than on time.
326fn compression() -> Option<i32> {
327    match std::env::var("LFSX_COMPRESSION").ok()?.trim() {
328        "" | "none" | "off" => None,
329        "zstd" => Some(3),
330        other => match other
331            .strip_prefix("zstd:")
332            .and_then(|level| level.parse().ok())
333        {
334            Some(level @ 1..=19) => Some(level),
335            _ => {
336                tracing::warn!(
337                    "LFSX_COMPRESSION={other} is not a codec this server knows — storing objects as they arrive"
338                );
339                None
340            }
341        },
342    }
343}
344
345fn bytes(variable: &str) -> Option<u64> {
346    let configured = std::env::var(variable).ok()?.trim().parse().ok()?;
347
348    if configured == 0 {
349        tracing::warn!("{variable}=0 would refuse every upload — ignoring it");
350        return None;
351    }
352
353    Some(configured)
354}
355
356fn seconds(variable: &str) -> Option<Duration> {
357    std::env::var(variable)
358        .ok()
359        .and_then(|raw| raw.parse().ok())
360        .map(Duration::from_secs)
361}
362
363#[cfg(test)]
364mod tests;
lfsx_server/config.rs

lfsx_server/
config.rs
