TLA Line data Source code
1 : // Download extension files from the extension store
2 : // and put them in the right place in the postgres directory (share / lib)
3 : /*
4 : The layout of the S3 bucket is as follows:
5 : 5615610098 // this is an extension build number
6 : ├── v14
7 : │ ├── extensions
8 : │ │ ├── anon.tar.zst
9 : │ │ └── embedding.tar.zst
10 : │ └── ext_index.json
11 : └── v15
12 : ├── extensions
13 : │ ├── anon.tar.zst
14 : │ └── embedding.tar.zst
15 : └── ext_index.json
16 : 5615261079
17 : ├── v14
18 : │ ├── extensions
19 : │ │ └── anon.tar.zst
20 : │ └── ext_index.json
21 : └── v15
22 : ├── extensions
23 : │ └── anon.tar.zst
24 : └── ext_index.json
25 : 5623261088
26 : ├── v14
27 : │ ├── extensions
28 : │ │ └── embedding.tar.zst
29 : │ └── ext_index.json
30 : └── v15
31 : ├── extensions
32 : │ └── embedding.tar.zst
33 : └── ext_index.json
34 :
35 : Note that build number cannot be part of prefix because we might need extensions
36 : from other build numbers.
37 :
38 : ext_index.json stores the control files and location of extension archives
39 : It also stores a list of public extensions and a library_index
40 :
41 : We don't need to duplicate extension.tar.zst files.
42 : We only need to upload a new one if it is updated.
43 : (Although currently we just upload every time anyways, hopefully will change
44 : this sometime)
45 :
46 : *access* is controlled by spec
47 :
48 : More specifically, here is an example ext_index.json
49 : {
50 : "public_extensions": [
51 : "anon",
52 : "pg_buffercache"
53 : ],
54 : "library_index": {
55 : "anon": "anon",
56 : "pg_buffercache": "pg_buffercache"
57 : },
58 : "extension_data": {
59 : "pg_buffercache": {
60 : "control_data": {
61 : "pg_buffercache.control": "# pg_buffercache extension \ncomment = 'examine the shared buffer cache' \ndefault_version = '1.3' \nmodule_pathname = '$libdir/pg_buffercache' \nrelocatable = true \ntrusted=true"
62 : },
63 : "archive_path": "5670669815/v14/extensions/pg_buffercache.tar.zst"
64 : },
65 : "anon": {
66 : "control_data": {
67 : "anon.control": "# PostgreSQL Anonymizer (anon) extension \ncomment = 'Data anonymization tools' \ndefault_version = '1.1.0' \ndirectory='extension/anon' \nrelocatable = false \nrequires = 'pgcrypto' \nsuperuser = false \nmodule_pathname = '$libdir/anon' \ntrusted = true \n"
68 : },
69 : "archive_path": "5670669815/v14/extensions/anon.tar.zst"
70 : }
71 : }
72 : }
73 : */
74 : use anyhow::Context;
75 : use anyhow::{self, Result};
76 : use compute_api::spec::RemoteExtSpec;
77 : use regex::Regex;
78 : use remote_storage::*;
79 : use serde_json;
80 : use std::io::Read;
81 : use std::num::{NonZeroU32, NonZeroUsize};
82 : use std::path::Path;
83 : use std::str;
84 : use tar::Archive;
85 : use tokio::io::AsyncReadExt;
86 : use tracing::info;
87 : use tracing::log::warn;
88 : use zstd::stream::read::Decoder;
89 :
90 CBC 641 : fn get_pg_config(argument: &str, pgbin: &str) -> String {
91 641 : // gives the result of `pg_config [argument]`
92 641 : // where argument is a flag like `--version` or `--sharedir`
93 641 : let pgconfig = pgbin
94 641 : .strip_suffix("postgres")
95 641 : .expect("bad pgbin")
96 641 : .to_owned()
97 641 : + "/pg_config";
98 641 : let config_output = std::process::Command::new(pgconfig)
99 641 : .arg(argument)
100 641 : .output()
101 641 : .expect("pg_config error");
102 641 : std::str::from_utf8(&config_output.stdout)
103 641 : .expect("pg_config error")
104 641 : .trim()
105 641 : .to_string()
106 641 : }
107 :
108 641 : pub fn get_pg_version(pgbin: &str) -> String {
109 641 : // pg_config --version returns a (platform specific) human readable string
110 641 : // such as "PostgreSQL 15.4". We parse this to v14/v15/v16 etc.
111 641 : let human_version = get_pg_config("--version", pgbin);
112 641 : return parse_pg_version(&human_version).to_string();
113 641 : }
114 :
115 653 : fn parse_pg_version(human_version: &str) -> &str {
116 653 : // Normal releases have version strings like "PostgreSQL 15.4". But there
117 653 : // are also pre-release versions like "PostgreSQL 17devel" or "PostgreSQL
118 653 : // 16beta2" or "PostgreSQL 17rc1". And with the --with-extra-version
119 653 : // configure option, you can tack any string to the version number,
120 653 : // e.g. "PostgreSQL 15.4foobar".
121 653 : match Regex::new(r"^PostgreSQL (?<major>\d+).+")
122 653 : .unwrap()
123 653 : .captures(human_version)
124 : {
125 653 : Some(captures) if captures.len() == 2 => match &captures["major"] {
126 653 : "14" => return "v14",
127 9 : "15" => return "v15",
128 6 : "16" => return "v16",
129 2 : _ => {}
130 : },
131 UBC 0 : _ => {}
132 : }
133 CBC 2 : panic!("Unsuported postgres version {human_version}");
134 651 : }
135 :
136 : #[cfg(test)]
137 : mod tests {
138 : use super::parse_pg_version;
139 :
140 1 : #[test]
141 1 : fn test_parse_pg_version() {
142 1 : assert_eq!(parse_pg_version("PostgreSQL 15.4"), "v15");
143 1 : assert_eq!(parse_pg_version("PostgreSQL 15.14"), "v15");
144 1 : assert_eq!(
145 1 : parse_pg_version("PostgreSQL 15.4 (Ubuntu 15.4-0ubuntu0.23.04.1)"),
146 1 : "v15"
147 1 : );
148 :
149 1 : assert_eq!(parse_pg_version("PostgreSQL 14.15"), "v14");
150 1 : assert_eq!(parse_pg_version("PostgreSQL 14.0"), "v14");
151 1 : assert_eq!(
152 1 : parse_pg_version("PostgreSQL 14.9 (Debian 14.9-1.pgdg120+1"),
153 1 : "v14"
154 1 : );
155 :
156 1 : assert_eq!(parse_pg_version("PostgreSQL 16devel"), "v16");
157 1 : assert_eq!(parse_pg_version("PostgreSQL 16beta1"), "v16");
158 1 : assert_eq!(parse_pg_version("PostgreSQL 16rc2"), "v16");
159 1 : assert_eq!(parse_pg_version("PostgreSQL 16extra"), "v16");
160 1 : }
161 :
162 1 : #[test]
163 : #[should_panic]
164 1 : fn test_parse_pg_unsupported_version() {
165 1 : parse_pg_version("PostgreSQL 13.14");
166 1 : }
167 :
168 1 : #[test]
169 : #[should_panic]
170 1 : fn test_parse_pg_incorrect_version_format() {
171 1 : parse_pg_version("PostgreSQL 14");
172 1 : }
173 : }
174 :
175 : // download the archive for a given extension,
176 : // unzip it, and place files in the appropriate locations (share/lib)
177 UBC 0 : pub async fn download_extension(
178 0 : ext_name: &str,
179 0 : ext_path: &RemotePath,
180 0 : remote_storage: &GenericRemoteStorage,
181 0 : pgbin: &str,
182 0 : ) -> Result<u64> {
183 0 : info!("Download extension {:?} from {:?}", ext_name, ext_path);
184 0 : let mut download = remote_storage.download(ext_path).await?;
185 0 : let mut download_buffer = Vec::new();
186 0 : download
187 0 : .download_stream
188 0 : .read_to_end(&mut download_buffer)
189 0 : .await?;
190 0 : let download_size = download_buffer.len() as u64;
191 : // it's unclear whether it is more performant to decompress into memory or not
192 : // TODO: decompressing into memory can be avoided
193 0 : let mut decoder = Decoder::new(download_buffer.as_slice())?;
194 0 : let mut decompress_buffer = Vec::new();
195 0 : decoder.read_to_end(&mut decompress_buffer)?;
196 0 : let mut archive = Archive::new(decompress_buffer.as_slice());
197 0 : let unzip_dest = pgbin
198 0 : .strip_suffix("/bin/postgres")
199 0 : .expect("bad pgbin")
200 0 : .to_string()
201 0 : + "/download_extensions";
202 0 : archive.unpack(&unzip_dest)?;
203 0 : info!("Download + unzip {:?} completed successfully", &ext_path);
204 :
205 0 : let sharedir_paths = (
206 0 : unzip_dest.to_string() + "/share/extension",
207 0 : Path::new(&get_pg_config("--sharedir", pgbin)).join("extension"),
208 0 : );
209 0 : let libdir_paths = (
210 0 : unzip_dest.to_string() + "/lib",
211 0 : Path::new(&get_pg_config("--pkglibdir", pgbin)).to_path_buf(),
212 0 : );
213 : // move contents of the libdir / sharedir in unzipped archive to the correct local paths
214 0 : for paths in [sharedir_paths, libdir_paths] {
215 0 : let (zip_dir, real_dir) = paths;
216 0 : info!("mv {zip_dir:?}/* {real_dir:?}");
217 0 : for file in std::fs::read_dir(zip_dir)? {
218 0 : let old_file = file?.path();
219 0 : let new_file =
220 0 : Path::new(&real_dir).join(old_file.file_name().context("error parsing file")?);
221 0 : info!("moving {old_file:?} to {new_file:?}");
222 :
223 : // extension download failed: Directory not empty (os error 39)
224 0 : match std::fs::rename(old_file, new_file) {
225 0 : Ok(()) => info!("move succeeded"),
226 0 : Err(e) => {
227 0 : warn!("move failed, probably because the extension already exists: {e}")
228 : }
229 : }
230 : }
231 : }
232 0 : info!("done moving extension {ext_name}");
233 0 : Ok(download_size)
234 0 : }
235 :
236 : // Create extension control files from spec
237 0 : pub fn create_control_files(remote_extensions: &RemoteExtSpec, pgbin: &str) {
238 0 : let local_sharedir = Path::new(&get_pg_config("--sharedir", pgbin)).join("extension");
239 0 : for (ext_name, ext_data) in remote_extensions.extension_data.iter() {
240 : // Check if extension is present in public or custom.
241 : // If not, then it is not allowed to be used by this compute.
242 0 : if let Some(public_extensions) = &remote_extensions.public_extensions {
243 0 : if !public_extensions.contains(ext_name) {
244 0 : if let Some(custom_extensions) = &remote_extensions.custom_extensions {
245 0 : if !custom_extensions.contains(ext_name) {
246 0 : continue; // skip this extension, it is not allowed
247 0 : }
248 0 : }
249 0 : }
250 0 : }
251 :
252 0 : for (control_name, control_content) in &ext_data.control_data {
253 0 : let control_path = local_sharedir.join(control_name);
254 0 : if !control_path.exists() {
255 0 : info!("writing file {:?}{:?}", control_path, control_content);
256 0 : std::fs::write(control_path, control_content).unwrap();
257 : } else {
258 0 : warn!("control file {:?} exists both locally and remotely. ignoring the remote version.", control_path);
259 : }
260 : }
261 : }
262 0 : }
263 :
264 : // This function initializes the necessary structs to use remote storage
265 0 : pub fn init_remote_storage(remote_ext_config: &str) -> anyhow::Result<GenericRemoteStorage> {
266 0 : #[derive(Debug, serde::Deserialize)]
267 : struct RemoteExtJson {
268 : bucket: String,
269 : region: String,
270 : endpoint: Option<String>,
271 : prefix: Option<String>,
272 : }
273 0 : let remote_ext_json = serde_json::from_str::<RemoteExtJson>(remote_ext_config)?;
274 :
275 0 : let config = S3Config {
276 0 : bucket_name: remote_ext_json.bucket,
277 0 : bucket_region: remote_ext_json.region,
278 0 : prefix_in_bucket: remote_ext_json.prefix,
279 0 : endpoint: remote_ext_json.endpoint,
280 0 : concurrency_limit: NonZeroUsize::new(100).expect("100 != 0"),
281 0 : max_keys_per_list_response: None,
282 0 : };
283 0 : let config = RemoteStorageConfig {
284 0 : max_concurrent_syncs: NonZeroUsize::new(100).expect("100 != 0"),
285 0 : max_sync_errors: NonZeroU32::new(100).expect("100 != 0"),
286 0 : storage: RemoteStorageKind::AwsS3(config),
287 0 : };
288 0 : GenericRemoteStorage::from_config(&config)
289 0 : }
|
