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    },
53}
54
55impl Storage {
56    fn from_env() -> Self {
57        if std::env::var("LFSX_STORAGE").as_deref() != Ok("s3") {
58            return Self::Local;
59        }
60
61        let required = |name: &str| {
62            std::env::var(name)
63                .ok()
64                .filter(|value| !value.is_empty())
65                .unwrap_or_else(|| panic!("LFSX_STORAGE=s3 needs {name}"))
66        };
67
68        Self::Bucket {
69            endpoint: required("LFSX_S3_ENDPOINT"),
70            bucket: required("LFSX_S3_BUCKET"),
71            region: std::env::var("LFSX_S3_REGION").unwrap_or_else(|_| "us-east-1".into()),
72            access_key: required("LFSX_S3_ACCESS_KEY"),
73            secret_key: required("LFSX_S3_SECRET_KEY"),
74            path_style: std::env::var("LFSX_S3_PATH_STYLE").as_deref() != Ok("false"),
75            presign: std::env::var("LFSX_S3_PRESIGN").as_deref() == Ok("true"),
76        }
77    }
78}
79
80#[derive(Debug, Clone)]
81pub enum Auth {
82    Forge {
83        provider: Provider,
84        api_url: String,
85        cache_ttl: Duration,
86        rejection_ttl: Duration,
87        // Whether a request with no credentials is resolved against the forge
88        // instead of refused. On by default, because that is what cloning a
89        // public repository does everywhere else.
90        anonymous_read: bool,
91    },
92    Disabled,
93}
94
95#[derive(Debug, Clone, Copy, PartialEq, Eq)]
96pub enum Provider {
97    Github,
98    Gitlab,
99    // Gitea and Forgejo, which are one API: Forgejo is a fork of Gitea and
100    // answers the same routes, so the only thing that tells two instances apart
101    // is the root they are reached at.
102    Gitea,
103}
104
105impl Provider {
106    // None where there is no such thing as the instance.
107    //
108    // github.com and gitlab.com are where a repository is unless the operator
109    // says otherwise, so defaulting there is nearly always right. Gitea is
110    // software rather than a place. gitea.com exists, but an operator who names
111    // this provider is almost certainly running their own, and quietly resolving
112    // their namespaces against a stranger's forge is worse than not starting: a
113    // public repository there that happens to share a name would hand out
114    // anonymous read on objects it has nothing to do with.
115    fn default_api_url(self) -> Option<&'static str> {
116        match self {
117            Self::Github => Some("https://api.github.com"),
118            Self::Gitlab => Some("https://gitlab.com/api/v4"),
119            Self::Gitea => None,
120        }
121    }
122
123    fn api_url_variable(self) -> &'static str {
124        match self {
125            Self::Github => "LFSX_GITHUB_API_URL",
126            Self::Gitlab => "LFSX_GITLAB_API_URL",
127            Self::Gitea => "LFSX_GITEA_API_URL",
128        }
129    }
130}
131
132const CACHE_TTL: Duration = Duration::from_secs(60);
133const REJECTION_TTL: Duration = Duration::from_secs(10);
134const GC_GRACE: Duration = Duration::from_secs(14 * 24 * 60 * 60);
135const STAGING_MAX_AGE: Duration = Duration::from_secs(24 * 60 * 60);
136
137impl Config {
138    pub fn from_env() -> Self {
139        let bind = std::env::var("LFSX_BIND")
140            .ok()
141            .and_then(|raw| raw.parse().ok())
142            .unwrap_or_else(|| SocketAddr::from(([0, 0, 0, 0], 8080)));
143
144        let storage_root = std::env::var("LFSX_STORAGE_ROOT")
145            .map(PathBuf::from)
146            .unwrap_or_else(|_| PathBuf::from("/var/lib/lfsx"));
147
148        let public_url = std::env::var("LFSX_PUBLIC_URL")
149            .ok()
150            .filter(|url| !url.is_empty())
151            .map(|url| url.trim_end_matches('/').to_owned());
152
153        Self {
154            bind,
155            storage_root,
156            public_url,
157            action_lifetime: 1800,
158            gc_grace: seconds("LFSX_GC_GRACE").unwrap_or(GC_GRACE),
159            staging_max_age: seconds("LFSX_STAGING_MAX_AGE").unwrap_or(STAGING_MAX_AGE),
160            lock_max_age: seconds("LFSX_LOCK_MAX_AGE"),
161            max_object_size: bytes("LFSX_MAX_OBJECT_SIZE"),
162            repo_quota: bytes("LFSX_REPO_QUOTA"),
163            compression: compression(),
164            encryption_key_file: std::env::var("LFSX_ENCRYPTION_KEY_FILE")
165                .ok()
166                .filter(|path| !path.is_empty())
167                .map(PathBuf::from),
168            storage: Storage::from_env(),
169            auth: Auth::from_env(),
170        }
171    }
172
173    pub fn base_url(&self, headers: &HeaderMap) -> String {
174        if let Some(configured) = &self.public_url {
175            return configured.clone();
176        }
177
178        let scheme = headers
179            .get("x-forwarded-proto")
180            .and_then(|value| value.to_str().ok())
181            .and_then(|value| value.split(',').next())
182            .map(str::trim)
183            .filter(|scheme| !scheme.is_empty())
184            .unwrap_or("http");
185
186        let authority = headers
187            .get(header::HOST)
188            .and_then(|value| value.to_str().ok())
189            .map(str::trim)
190            .filter(|host| !host.is_empty())
191            .unwrap_or("localhost");
192
193        format!("{scheme}://{authority}")
194    }
195
196    pub fn object_url(&self, base: &str, ns: &Namespace, oid: &str) -> String {
197        format!("{base}/{ns}/objects/{oid}")
198    }
199
200    pub fn verify_url(&self, base: &str, ns: &Namespace) -> String {
201        format!("{base}/{ns}/objects/verify")
202    }
203
204    pub fn action(&self, href: String) -> Action {
205        Action {
206            href,
207            header: None,
208            expires_in: self.action_lifetime,
209        }
210    }
211
212    pub fn signed_action(&self, href: String, headers: Vec<(String, String)>) -> Action {
213        Action {
214            href,
215            header: Some(headers.into_iter().collect()),
216            expires_in: self.action_lifetime,
217        }
218    }
219}
220
221// Opt in, not opt out. Serving objects to a caller with no credentials at all is
222// a decision an operator should make on purpose: it costs them the bandwidth of
223// anyone who finds the endpoint, on a server whose whole job is to move files
224// measured in gigabytes. Nothing confidential is at stake, since a request with
225// no credentials is still resolved against the forge and a private repository is
226// still refused, but "anyone may pull from you" is not a sensible thing to
227// inherit by default.
228//
229// Only the exact string opens it. A typo, an empty value or a `1` leaves it
230// closed, because the failure that matters here is the one that opens the door
231// when nobody meant to.
232fn anonymous_read(value: Option<&str>) -> bool {
233    value == Some("true")
234}
235
236impl Auth {
237    fn from_env() -> Self {
238        if std::env::var("LFSX_AUTH").as_deref() == Ok("disabled") {
239            tracing::warn!(
240                "LFSX_AUTH=disabled — every request is accepted, run this on a trusted network only"
241            );
242            return Self::Disabled;
243        }
244
245        let provider = provider(std::env::var("LFSX_AUTH").ok().as_deref());
246
247        Self::Forge {
248            provider,
249            api_url: api_url(
250                provider,
251                std::env::var(provider.api_url_variable()).ok().as_deref(),
252            ),
253            cache_ttl: seconds("LFSX_AUTH_CACHE_TTL").unwrap_or(CACHE_TTL),
254            rejection_ttl: seconds("LFSX_AUTH_REJECTION_TTL").unwrap_or(REJECTION_TTL),
255            anonymous_read: anonymous_read(std::env::var("LFSX_ANONYMOUS_READ").ok().as_deref()),
256        }
257    }
258}
259
260// Anything unrecognised is GitHub, which is what an operator who set nothing
261// almost certainly meant. Forgejo is named alongside Gitea because they are one
262// API, and somebody running Forgejo should not have to know it began as a fork.
263fn provider(value: Option<&str>) -> Provider {
264    match value {
265        Some("gitlab") => Provider::Gitlab,
266        Some("gitea") | Some("forgejo") => Provider::Gitea,
267        _ => Provider::Github,
268    }
269}
270
271// The trailing slash matters: every route is built by appending to this, so one
272// left on the end produces `//repos/...`, which some forges answer and others do
273// not, and the ones that do not answer 404 for a repository that is right there.
274fn api_url(provider: Provider, configured: Option<&str>) -> String {
275    let variable = provider.api_url_variable();
276
277    configured
278        .map(str::to_owned)
279        .or_else(|| provider.default_api_url().map(str::to_owned))
280        .unwrap_or_else(|| {
281            panic!(
282                "{variable} must be set: a self-hosted forge has no default API root, and guessing \
283                 one would resolve your repositories against somebody else's"
284            )
285        })
286        .trim_end_matches('/')
287        .to_owned()
288}
289
290// Unset means unlimited, which is what a server on its own volume wants. Zero
291// would refuse every push, so it is read as a typo rather than as a policy
292// nobody would choose deliberately.
293// zstd level 3 is the default because it is the one that costs nothing you can
294// measure: it compresses faster than a spinning disk writes, and the meshes and
295// uncompressed raster that make up most of an LFS store give most of their
296// ground at any level. Higher levels are there for a store that is short on
297// room rather than on time.
298fn compression() -> Option<i32> {
299    match std::env::var("LFSX_COMPRESSION").ok()?.trim() {
300        "" | "none" | "off" => None,
301        "zstd" => Some(3),
302        other => match other
303            .strip_prefix("zstd:")
304            .and_then(|level| level.parse().ok())
305        {
306            Some(level @ 1..=19) => Some(level),
307            _ => {
308                tracing::warn!(
309                    "LFSX_COMPRESSION={other} is not a codec this server knows — storing objects as they arrive"
310                );
311                None
312            }
313        },
314    }
315}
316
317fn bytes(variable: &str) -> Option<u64> {
318    let configured = std::env::var(variable).ok()?.trim().parse().ok()?;
319
320    if configured == 0 {
321        tracing::warn!("{variable}=0 would refuse every upload — ignoring it");
322        return None;
323    }
324
325    Some(configured)
326}
327
328fn seconds(variable: &str) -> Option<Duration> {
329    std::env::var(variable)
330        .ok()
331        .and_then(|raw| raw.parse().ok())
332        .map(Duration::from_secs)
333}
334
335#[cfg(test)]
336mod tests;
lfsx_server/config.rs

lfsx_server/
config.rs
