Add trailing dot

Conflicts:
	control_plane/src/pageserver.rs
    => pick ours

Cargo.lock

@@ -1348,6 +1348,7 @@ dependencies = [
  "tokio-postgres",
  "tokio-util",
  "toml",
+ "toml_edit",
  "tracing",
  "url",
  "utils",

control_plane/Cargo.toml

@@ -28,6 +28,7 @@ serde_with.workspace = true
 tar.workspace = true
 thiserror.workspace = true
 toml.workspace = true
+toml_edit.workspace = true
 tokio.workspace = true
 tokio-postgres.workspace = true
 tokio-util.workspace = true

control_plane/src/bin/neon_local.rs

@@ -133,7 +133,7 @@ fn main() -> Result<()> {
         let subcommand_result = match sub_name {
             "tenant" => rt.block_on(handle_tenant(sub_args, &mut env)),
             "timeline" => rt.block_on(handle_timeline(sub_args, &mut env)),
-            "start" => rt.block_on(handle_start_all(sub_args, &env)),
+            "start" => rt.block_on(handle_start_all(&env)),
             "stop" => rt.block_on(handle_stop_all(sub_args, &env)),
             "pageserver" => rt.block_on(handle_pageserver(sub_args, &env)),
             "storage_controller" => rt.block_on(handle_storage_controller(sub_args, &env)),
@@ -358,6 +358,13 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
         default_conf(*num_pageservers)
     };
 
+    let pageserver_config: toml_edit::Document =
+        if let Some(path) = init_match.get_one::<PathBuf>("pageserver-config") {
+            std::fs::read_to_string(path)?.parse()?
+        } else {
+            toml_edit::Document::new()
+        };
+
     let pg_version = init_match
         .get_one::<u32>("pg-version")
         .copied()
@@ -375,7 +382,7 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
     // Initialize pageserver, create initial tenant and timeline.
     for ps_conf in &env.pageservers {
         PageServerNode::from_env(&env, ps_conf)
-            .initialize(&pageserver_config_overrides(init_match))
+            .initialize(&pageserver_config)
             .unwrap_or_else(|e| {
                 eprintln!("pageserver init failed: {e:?}");
                 exit(1);
@@ -397,15 +404,6 @@ fn get_default_pageserver(env: &local_env::LocalEnv) -> PageServerNode {
     PageServerNode::from_env(env, ps_conf)
 }
 
-fn pageserver_config_overrides(init_match: &ArgMatches) -> Vec<&str> {
-    init_match
-        .get_many::<String>("pageserver-config-override")
-        .into_iter()
-        .flatten()
-        .map(String::as_str)
-        .collect()
-}
-
 async fn handle_tenant(
     tenant_match: &ArgMatches,
     env: &mut local_env::LocalEnv,
@@ -1068,10 +1066,7 @@ fn get_pageserver(env: &local_env::LocalEnv, args: &ArgMatches) -> Result<PageSe
 async fn handle_pageserver(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
     match sub_match.subcommand() {
         Some(("start", subcommand_args)) => {
-            if let Err(e) = get_pageserver(env, subcommand_args)?
+            if let Err(e) = get_pageserver(env, subcommand_args)?.start().await {
-                .start(&pageserver_config_overrides(subcommand_args))
-                .await
-            {
                 eprintln!("pageserver start failed: {e}");
                 exit(1);
             }
@@ -1097,10 +1092,7 @@ async fn handle_pageserver(sub_match: &ArgMatches, env: &local_env::LocalEnv) ->
                 exit(1);
             }
 
-            if let Err(e) = pageserver
+            if let Err(e) = pageserver.start().await {
-                .start(&pageserver_config_overrides(subcommand_args))
-                .await
-            {
                 eprintln!("pageserver start failed: {e}");
                 exit(1);
             }
@@ -1227,7 +1219,7 @@ async fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) ->
     Ok(())
 }
 
-async fn handle_start_all(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> anyhow::Result<()> {
+async fn handle_start_all(env: &local_env::LocalEnv) -> anyhow::Result<()> {
     // Endpoints are not started automatically
 
     broker::start_broker_process(env).await?;
@@ -1244,10 +1236,7 @@ async fn handle_start_all(sub_match: &ArgMatches, env: &local_env::LocalEnv) ->
 
     for ps_conf in &env.pageservers {
         let pageserver = PageServerNode::from_env(env, ps_conf);
-        if let Err(e) = pageserver
+        if let Err(e) = pageserver.start().await {
-            .start(&pageserver_config_overrides(sub_match))
-            .await
-        {
             eprintln!("pageserver {} start failed: {:#}", ps_conf.id, e);
             try_stop_all(env, true).await;
             exit(1);
@@ -1388,13 +1377,6 @@ fn cli() -> Command {
         .required(false)
         .value_name("stop-mode");
 
-    let pageserver_config_args = Arg::new("pageserver-config-override")
-        .long("pageserver-config-override")
-        .num_args(1)
-        .action(ArgAction::Append)
-        .help("Additional pageserver's configuration options or overrides, refer to pageserver's 'config-override' CLI parameter docs for more")
-        .required(false);
-
     let remote_ext_config_args = Arg::new("remote-ext-config")
         .long("remote-ext-config")
         .num_args(1)
@@ -1450,14 +1432,21 @@ fn cli() -> Command {
         .subcommand(
             Command::new("init")
                 .about("Initialize a new Neon repository, preparing configs for services to start with")
-                .arg(pageserver_config_args.clone())
                 .arg(num_pageservers_arg.clone())
                 .arg(
                     Arg::new("config")
                         .long("config")
                         .required(false)
                         .value_parser(value_parser!(PathBuf))
-                        .value_name("config"),
+                        .value_name("config")
+                )
+                .arg(
+                    Arg::new("pageserver-config")
+                        .long("pageserver-config")
+                        .required(false)
+                        .value_parser(value_parser!(PathBuf))
+                        .value_name("pageserver-config")
+                        .help("Merge the provided pageserver config into the one generated by neon_local."),
                 )
                 .arg(pg_version_arg.clone())
                 .arg(force_arg)
@@ -1539,7 +1528,6 @@ fn cli() -> Command {
                 .subcommand(Command::new("status"))
                 .subcommand(Command::new("start")
                     .about("Start local pageserver")
-                    .arg(pageserver_config_args.clone())
                 )
                 .subcommand(Command::new("stop")
                     .about("Stop local pageserver")
@@ -1547,7 +1535,6 @@ fn cli() -> Command {
                 )
                 .subcommand(Command::new("restart")
                     .about("Restart local pageserver")
-                    .arg(pageserver_config_args.clone())
                 )
         )
         .subcommand(
@@ -1660,7 +1647,6 @@ fn cli() -> Command {
         .subcommand(
             Command::new("start")
                 .about("Start page server and safekeepers")
-                .arg(pageserver_config_args)
         )
         .subcommand(
             Command::new("stop")

control_plane/src/pageserver.rs

@@ -4,7 +4,6 @@
 //!
 //!   .neon/
 //!
-use std::borrow::Cow;
 use std::collections::HashMap;
 
 use std::io;
@@ -77,7 +76,7 @@ impl PageServerNode {
     /// Merge overrides provided by the user on the command line with our default overides derived from neon_local configuration.
     ///
     /// These all end up on the command line of the `pageserver` binary.
-    fn neon_local_overrides(&self, cli_overrides: &[&str]) -> Vec<String> {
+    fn neon_local_overrides(&self, cli_overrides: &toml_edit::Document) -> Vec<String> {
         // FIXME: the paths should be shell-escaped to handle paths with spaces, quotas etc.
         let pg_distrib_dir_param = format!(
             "pg_distrib_dir='{}'",
@@ -157,10 +156,7 @@ impl PageServerNode {
             }
         }
 
-        if !cli_overrides
+        if !cli_overrides.contains_key("remote_storage") {
-            .iter()
-            .any(|c| c.starts_with("remote_storage"))
-        {
             overrides.push(format!(
                 "remote_storage={{local_path='../{PAGESERVER_REMOTE_STORAGE_DIR}'}}"
             ));
@@ -173,13 +169,13 @@ impl PageServerNode {
         }
 
         // Apply the user-provided overrides
-        overrides.extend(cli_overrides.iter().map(|&c| c.to_owned()));
+        overrides.push(cli_overrides.to_string());
 
         overrides
     }
 
     /// Initializes a pageserver node by creating its config with the overrides provided.
-    pub fn initialize(&self, config_overrides: &[&str]) -> anyhow::Result<()> {
+    pub fn initialize(&self, config_overrides: &toml_edit::Document) -> anyhow::Result<()> {
         // First, run `pageserver --init` and wait for it to write a config into FS and exit.
         self.pageserver_init(config_overrides)
             .with_context(|| format!("Failed to run init for pageserver node {}", self.conf.id))
@@ -197,11 +193,11 @@ impl PageServerNode {
             .expect("non-Unicode path")
     }
 
-    pub async fn start(&self, config_overrides: &[&str]) -> anyhow::Result<()> {
+    pub async fn start(&self) -> anyhow::Result<()> {
-        self.start_node(config_overrides, false).await
+        self.start_node().await
     }
 
-    fn pageserver_init(&self, config_overrides: &[&str]) -> anyhow::Result<()> {
+    fn pageserver_init(&self, config_overrides: &toml_edit::Document) -> anyhow::Result<()> {
         let datadir = self.repo_path();
         let node_id = self.conf.id;
         println!(
@@ -219,11 +215,18 @@ impl PageServerNode {
         let datadir_path_str = datadir.to_str().with_context(|| {
             format!("Cannot start pageserver node {node_id} in path that has no string representation: {datadir:?}")
         })?;
-        let mut args = self.pageserver_basic_args(config_overrides, datadir_path_str);
-        args.push(Cow::Borrowed("--init"));
 
+        // `pageserver --init` merges the `--config-override`s into a built-in default config,
+        // then writes out the merged product to `pageserver.toml`.
+        // TODO: just write the full `pageserver.toml` and get rid of `--config-override`.
+        let mut args = vec!["--init", "--workdir", datadir_path_str];
+        let overrides = self.neon_local_overrides(config_overrides);
+        for piece in &overrides {
+            args.push("--config-override");
+            args.push(piece);
+        }
         let init_output = Command::new(self.env.pageserver_bin())
-            .args(args.iter().map(Cow::as_ref))
+            .args(args)
             .envs(self.pageserver_env_variables()?)
             .output()
             .with_context(|| format!("Failed to run pageserver init for node {node_id}"))?;
@@ -262,11 +265,7 @@ impl PageServerNode {
         Ok(())
     }
 
-    async fn start_node(
+    async fn start_node(&self) -> anyhow::Result<()> {
-        &self,
-        config_overrides: &[&str],
-        update_config: bool,
-    ) -> anyhow::Result<()> {
         // TODO: using a thread here because start_process() is not async but we need to call check_status()
         let datadir = self.repo_path();
         print!(
@@ -283,15 +282,12 @@ impl PageServerNode {
                 self.conf.id, datadir,
             )
         })?;
-        let mut args = self.pageserver_basic_args(config_overrides, datadir_path_str);
+        let args = vec!["-D", datadir_path_str];
-        if update_config {
-            args.push(Cow::Borrowed("--update-config"));
-        }
         background_process::start_process(
             "pageserver",
             &datadir,
             &self.env.pageserver_bin(),
-            args.iter().map(Cow::as_ref),
+            args,
             self.pageserver_env_variables()?,
             background_process::InitialPidFile::Expect(self.pid_file()),
             || async {
@@ -308,22 +304,6 @@ impl PageServerNode {
         Ok(())
     }
 
-    fn pageserver_basic_args<'a>(
-        &self,
-        config_overrides: &'a [&'a str],
-        datadir_path_str: &'a str,
-    ) -> Vec<Cow<'a, str>> {
-        let mut args = vec![Cow::Borrowed("-D"), Cow::Borrowed(datadir_path_str)];
-
-        let overrides = self.neon_local_overrides(config_overrides);
-        for config_override in overrides {
-            args.push(Cow::Borrowed("-c"));
-            args.push(Cow::Owned(config_override));
-        }
-
-        args
-    }
-
     fn pageserver_env_variables(&self) -> anyhow::Result<Vec<(String, String)>> {
         // FIXME: why is this tied to pageserver's auth type? Whether or not the safekeeper
         // needs a token, and how to generate that token, seems independent to whether

pageserver/src/bin/pageserver.rs

@@ -3,6 +3,7 @@
 //! Main entry point for the Page Server executable.
 
 use std::env::{var, VarError};
+use std::io::Read;
 use std::sync::Arc;
 use std::time::Duration;
 use std::{env, ops::ControlFlow, str::FromStr};
@@ -151,37 +152,34 @@ fn initialize_config(
     workdir: &Utf8Path,
 ) -> anyhow::Result<ControlFlow<(), &'static PageServerConf>> {
     let init = arg_matches.get_flag("init");
-    let update_config = init || arg_matches.get_flag("update-config");
 
-    let (mut toml, config_file_exists) = if cfg_file_path.is_file() {
+    let file_contents: Option<toml_edit::Document> = match std::fs::File::open(cfg_file_path) {
-        if init {
+        Ok(mut f) => {
-            anyhow::bail!(
+            if init {
-                "Config file '{cfg_file_path}' already exists, cannot init it, use --update-config to update it",
+                anyhow::bail!("config file already exists: {cfg_file_path}");
-            );
+            }
+            let md = f.metadata().context("stat config file")?;
+            if md.is_file() {
+                let mut s = String::new();
+                f.read_to_string(&mut s).context("read config file")?;
+                Some(s.parse().context("parse config file toml")?)
+            } else {
+                anyhow::bail!("directory entry exists but is not a file: {cfg_file_path}");
+            }
+        }
+        Err(e) if e.kind() == std::io::ErrorKind::NotFound => None,
+        Err(e) => {
+            anyhow::bail!("open pageserver config: {e}: {cfg_file_path}");
         }
-        // Supplement the CLI arguments with the config file
-        let cfg_file_contents = std::fs::read_to_string(cfg_file_path)
-            .with_context(|| format!("Failed to read pageserver config at '{cfg_file_path}'"))?;
-        (
-            cfg_file_contents
-                .parse::<toml_edit::Document>()
-                .with_context(|| {
-                    format!("Failed to parse '{cfg_file_path}' as pageserver config")
-                })?,
-            true,
-        )
-    } else if cfg_file_path.exists() {
-        anyhow::bail!("Config file '{cfg_file_path}' exists but is not a regular file");
-    } else {
-        // We're initializing the tenant, so there's no config file yet
-        (
-            DEFAULT_CONFIG_FILE
-                .parse::<toml_edit::Document>()
-                .context("could not parse built-in config file")?,
-            false,
-        )
     };
 
+    let mut effective_config = file_contents.unwrap_or_else(|| {
+        DEFAULT_CONFIG_FILE
+            .parse()
+            .expect("unit tests ensure this works")
+    });
+
+    // Patch with overrides from the command line
     if let Some(values) = arg_matches.get_many::<String>("config-override") {
         for option_line in values {
             let doc = toml_edit::Document::from_str(option_line).with_context(|| {
@@ -189,22 +187,21 @@ fn initialize_config(
             })?;
 
             for (key, item) in doc.iter() {
-                if config_file_exists && update_config && key == "id" && toml.contains_key(key) {
+                effective_config.insert(key, item.clone());
-                    anyhow::bail!("Pageserver config file exists at '{cfg_file_path}' and has node id already, it cannot be overridden");
-                }
-                toml.insert(key, item.clone());
             }
         }
     }
 
-    debug!("Resulting toml: {toml}");
+    debug!("Resulting toml: {effective_config}");
-    let conf = PageServerConf::parse_and_validate(&toml, workdir)
+
+    // Construct the runtime representation
+    let conf = PageServerConf::parse_and_validate(&effective_config, workdir)
         .context("Failed to parse pageserver configuration")?;
 
-    if update_config {
+    if init {
         info!("Writing pageserver config to '{cfg_file_path}'");
 
-        std::fs::write(cfg_file_path, toml.to_string())
+        std::fs::write(cfg_file_path, effective_config.to_string())
             .with_context(|| format!("Failed to write pageserver config to '{cfg_file_path}'"))?;
         info!("Config successfully written to '{cfg_file_path}'")
     }
@@ -758,18 +755,13 @@ fn cli() -> Command {
         // See `settings.md` for more details on the extra configuration patameters pageserver can process
         .arg(
             Arg::new("config-override")
+                .long("config-override")
                 .short('c')
                 .num_args(1)
                 .action(ArgAction::Append)
                 .help("Additional configuration overrides of the ones from the toml config file (or new ones to add there). \
                 Any option has to be a valid toml document, example: `-c=\"foo='hey'\"` `-c=\"foo={value=1}\"`"),
         )
-        .arg(
-            Arg::new("update-config")
-                .long("update-config")
-                .action(ArgAction::SetTrue)
-                .help("Update the config file when started"),
-        )
         .arg(
             Arg::new("enabled-features")
                 .long("enabled-features")

test_runner/README.md

@@ -76,13 +76,10 @@ you can use `--pg-version` argument.
 `TEST_OUTPUT`: Set the directory where test state and test output files
 should go.
 `TEST_SHARED_FIXTURES`: Try to re-use a single pageserver for all the tests.
-`NEON_PAGESERVER_OVERRIDES`: add a `;`-separated set of configs that will be passed as
 `RUST_LOG`: logging configuration to pass into Neon CLI
 
 Useful parameters and commands:
 
-`--pageserver-config-override=${value}` `-c` values to pass into pageserver through neon_local cli
-
 `--preserve-database-files` to preserve pageserver (layer) and safekeer (segment) timeline files on disk
 after running a test suite. Such files might be large, so removed by default; but might be useful for debugging or creation of svg images with layer file contents.
 

test_runner/fixtures/neon_fixtures.py

@@ -14,7 +14,7 @@ import textwrap
 import threading
 import time
 import uuid
-from contextlib import closing, contextmanager
+from contextlib import ExitStack, closing, contextmanager
 from dataclasses import dataclass
 from datetime import datetime
 from enum import Enum
@@ -450,6 +450,7 @@ class NeonEnvBuilder:
         test_output_dir: Path,
         test_overlay_dir: Optional[Path] = None,
         pageserver_remote_storage: Optional[RemoteStorage] = None,
+        # toml that will be decomposed into `--config-override` flags during `pageserver --init`
         pageserver_config_override: Optional[str] = None,
         num_safekeepers: int = 1,
         num_pageservers: int = 1,
@@ -1021,7 +1022,6 @@ class NeonEnv:
         self.neon_local_binpath = config.neon_binpath
         self.pg_distrib_dir = config.pg_distrib_dir
         self.endpoint_counter = 0
-        self.pageserver_config_override = config.pageserver_config_override
         self.storage_controller_config = config.storage_controller_config
 
         # generate initial tenant ID here instead of letting 'neon init' generate it,
@@ -1131,7 +1131,11 @@ class NeonEnv:
             cfg["safekeepers"].append(sk_cfg)
 
         log.info(f"Config: {cfg}")
-        self.neon_cli.init(cfg, force=config.config_init_force)
+        self.neon_cli.init(
+            cfg,
+            force=config.config_init_force,
+            pageserver_config_override=config.pageserver_config_override,
+        )
 
     def start(self):
         # Storage controller starts first, so that pageserver /re-attach calls don't
@@ -1703,30 +1707,47 @@ class NeonCli(AbstractNeonCli):
         self,
         config: Dict[str, Any],
         force: Optional[str] = None,
+        pageserver_config_override: Optional[str] = None,
     ) -> "subprocess.CompletedProcess[str]":
-        with tempfile.NamedTemporaryFile(mode="w+") as tmp:
+        remote_storage = self.env.pageserver_remote_storage
-            tmp.write(toml.dumps(config))
-            tmp.flush()
 
-            cmd = ["init", f"--config={tmp.name}", "--pg-version", self.env.pg_version]
+        ps_config = {}
+        if remote_storage is not None:
+            remote_storage_toml_table = remote_storage_to_toml_inline_table(remote_storage)
+            ps_config["remote_storage"] = remote_storage_toml_table
+
+        if pageserver_config_override is not None:
+            for o in pageserver_config_override.split(";"):
+                override = toml.loads(o)
+                for key, value in override.items():
+                    ps_config[key] = value
+
+        with ExitStack() as stack:
+            ps_config_file = stack.enter_context(tempfile.NamedTemporaryFile(mode="w+"))
+            ps_config_file.write(toml.dumps(ps_config))
+            ps_config_file.flush()
+
+            neon_local_config = stack.enter_context(tempfile.NamedTemporaryFile(mode="w+"))
+            neon_local_config.write(toml.dumps(config))
+            neon_local_config.flush()
+
+            cmd = [
+                "init",
+                f"--config={neon_local_config.name}",
+                "--pg-version",
+                self.env.pg_version,
+                f"--pageserver-config={ps_config_file.name}",
+            ]
 
             if force is not None:
                 cmd.extend(["--force", force])
 
-            storage = self.env.pageserver_remote_storage
-
-            append_pageserver_param_overrides(
-                params_to_update=cmd,
-                remote_storage=storage,
-                pageserver_config_override=self.env.pageserver_config_override,
-            )
-
             s3_env_vars = None
-            if isinstance(storage, S3Storage):
+            if isinstance(remote_storage, S3Storage):
-                s3_env_vars = storage.access_env_vars()
+                s3_env_vars = remote_storage.access_env_vars()
             res = self.raw_cli(cmd, extra_env_vars=s3_env_vars)
             res.check_returncode()
-            return res
+        return res
 
     def storage_controller_start(self):
         cmd = ["storage_controller", "start"]
@@ -1741,16 +1762,10 @@ class NeonCli(AbstractNeonCli):
     def pageserver_start(
         self,
         id: int,
-        overrides: Tuple[str, ...] = (),
         extra_env_vars: Optional[Dict[str, str]] = None,
     ) -> "subprocess.CompletedProcess[str]":
-        start_args = ["pageserver", "start", f"--id={id}", *overrides]
+        start_args = ["pageserver", "start", f"--id={id}"]
         storage = self.env.pageserver_remote_storage
-        append_pageserver_param_overrides(
-            params_to_update=start_args,
-            remote_storage=storage,
-            pageserver_config_override=self.env.pageserver_config_override,
-        )
 
         if isinstance(storage, S3Storage):
             s3_env_vars = storage.access_env_vars()
@@ -2408,9 +2423,47 @@ class NeonPageserver(PgProtocol, LogUtils):
             return self.workdir / "tenants"
         return self.workdir / "tenants" / str(tenant_shard_id)
 
+    @property
+    def config_toml_path(self) -> Path:
+        return self.workdir / "pageserver.toml"
+
+    def edit_config_toml(self, edit_fn: Callable[[Dict[str, Any]], bool]):
+        """
+        Edit the pageserver's config toml file in place.
+
+        The `edit_fn` is to manipulate the dict, and if it returns True, the file will be written.
+        If it returns False, no changes are made to the file system.
+        """
+        path = self.config_toml_path
+        with open(path, "r") as f:
+            config = toml.load(f)
+        save = edit_fn(config)
+        if save:
+            with open(path, "w") as f:
+                toml.dump(config, f)
+
+    def patch_config_toml_nonrecursive(self, patch: Dict[str, Any]) -> Dict[str, Any]:
+        """
+        Non-recursively merge the given `patch` dict into the existing config toml, using `dict.update()`.
+        Returns the replaced values.
+        If there was no previous value, the key is mapped to None.
+        This allows to restore the original value by calling this method with the returned dict.
+        """
+        replacements = {}
+
+        def doit(config: Dict[str, Any]) -> bool:
+            while len(patch) > 0:
+                key, new = patch.popitem()
+                old = config.get(key, None)
+                config[key] = new
+                replacements[key] = old
+            return True
+
+        self.edit_config_toml(doit)
+        return replacements
+
     def start(
         self,
-        overrides: Tuple[str, ...] = (),
         extra_env_vars: Optional[Dict[str, str]] = None,
     ) -> "NeonPageserver":
         """
@@ -2420,9 +2473,7 @@ class NeonPageserver(PgProtocol, LogUtils):
         """
         assert self.running is False
 
-        self.env.neon_cli.pageserver_start(
+        self.env.neon_cli.pageserver_start(self.id, extra_env_vars=extra_env_vars)
-            self.id, overrides=overrides, extra_env_vars=extra_env_vars
-        )
         self.running = True
         return self
 
@@ -2585,33 +2636,6 @@ class NeonPageserver(PgProtocol, LogUtils):
         )
 
 
-def append_pageserver_param_overrides(
-    params_to_update: List[str],
-    remote_storage: Optional[RemoteStorage],
-    pageserver_config_override: Optional[str] = None,
-):
-    if remote_storage is not None:
-        remote_storage_toml_table = remote_storage_to_toml_inline_table(remote_storage)
-
-        params_to_update.append(
-            f"--pageserver-config-override=remote_storage={remote_storage_toml_table}"
-        )
-    else:
-        params_to_update.append('--pageserver-config-override=remote_storage=""')
-
-    env_overrides = os.getenv("NEON_PAGESERVER_OVERRIDES")
-    if env_overrides is not None:
-        params_to_update += [
-            f"--pageserver-config-override={o.strip()}" for o in env_overrides.split(";")
-        ]
-
-    if pageserver_config_override is not None:
-        params_to_update += [
-            f"--pageserver-config-override={o.strip()}"
-            for o in pageserver_config_override.split(";")
-        ]
-
-
 class PgBin:
     """A helper class for executing postgres binaries"""
 

test_runner/performance/test_branch_creation.py

@@ -140,10 +140,13 @@ def test_branch_creation_many(neon_compare: NeonCompare, n_branches: int, shape:
 
     # start without gc so we can time compaction with less noise; use shorter
     # period for compaction so it starts earlier
+    def patch_default_tenant_config(config):
+        config["compaction_period"] = "3s"
+        config["gc_period"] = "0s"
+        return True
+
+    env.pageserver.edit_config_toml(patch_default_tenant_config)
     env.pageserver.start(
-        overrides=(
-            "--pageserver-config-override=tenant_config={ compaction_period = '3s', gc_period = '0s' }",
-        ),
         # this does print more than we want, but the number should be comparable between runs
         extra_env_vars={
             "RUST_LOG": f"[compaction_loop{{tenant_id={env.initial_tenant}}}]=debug,info"

test_runner/regress/test_disk_usage_eviction.py

@@ -5,7 +5,6 @@ from dataclasses import dataclass
 from typing import Any, Dict, Iterable, Tuple
 
 import pytest
-import toml
 from fixtures.log_helper import log
 from fixtures.neon_fixtures import (
     NeonEnv,
@@ -45,17 +44,15 @@ def test_min_resident_size_override_handling(
         ps_http.set_tenant_config(tenant_id, {})
         assert_config(tenant_id, None, default_tenant_conf_value)
 
-    env.pageserver.stop()
     if config_level_override is not None:
-        env.pageserver.start(
+
-            overrides=(
+        def set_min_resident_size(config):
-                "--pageserver-config-override=tenant_config={ min_resident_size_override =  "
+            config["tenant_config"] = {"min_resident_size": config_level_override}
-                + str(config_level_override)
+            return True
-                + " }",
+
-            )
+        env.pageserver.edit_config_toml(set_min_resident_size)
-        )
+    env.pageserver.stop()
-    else:
+    env.pageserver.start()
-        env.pageserver.start()
 
     tenant_id, _ = env.neon_cli.create_tenant()
     assert_overrides(tenant_id, config_level_override)
@@ -164,34 +161,32 @@ class EvictionEnv:
         usage eviction task is unknown; it might need to run one more iteration
         before assertions can be made.
         """
-        disk_usage_config = {
-            "period": period,
-            "max_usage_pct": max_usage_pct,
-            "min_avail_bytes": min_avail_bytes,
-            "mock_statvfs": mock_behavior,
-            "eviction_order": eviction_order.config(),
-        }
-
-        enc = toml.TomlEncoder()
 
         # these can sometimes happen during startup before any tenants have been
         # loaded, so nothing can be evicted, we just wait for next iteration which
         # is able to evict.
         pageserver.allowed_errors.append(".*WARN.* disk usage still high.*")
 
-        pageserver.start(
+        pageserver.patch_config_toml_nonrecursive(
-            overrides=(
+            {
-                "--pageserver-config-override=disk_usage_based_eviction="
+                "disk_usage_based_eviction": {
-                + enc.dump_inline_table(disk_usage_config).replace("\n", " "),
+                    "period": period,
+                    "max_usage_pct": max_usage_pct,
+                    "min_avail_bytes": min_avail_bytes,
+                    "mock_statvfs": mock_behavior,
+                    "eviction_order": eviction_order.config(),
+                },
                 # Disk usage based eviction runs as a background task.
                 # But pageserver startup delays launch of background tasks for some time, to prioritize initial logical size calculations during startup.
                 # But, initial logical size calculation may not be triggered if safekeepers don't publish new broker messages.
                 # But, we only have a 10-second-timeout in this test.
                 # So, disable the delay for this test.
-                "--pageserver-config-override=background_task_maximum_delay='0s'",
+                "background_task_maximum_delay": "0s",
-            ),
+            }
         )
 
+        pageserver.start()
+
         # we now do initial logical size calculation on startup, which on debug builds can fight with disk usage based eviction
         for tenant_id, timeline_id in self.timelines:
             tenant_ps = self.neon_env.get_tenant_pageserver(tenant_id)

test_runner/regress/test_pageserver_api.py

@@ -12,7 +12,6 @@ from fixtures.types import Lsn, TenantId, TimelineId
 from fixtures.utils import wait_until
 
 
-# test that we cannot override node id after init
 def test_pageserver_init_node_id(
     neon_simple_env: NeonEnv, neon_binpath: Path, pg_distrib_dir: Path
 ):
@@ -49,11 +48,7 @@ def test_pageserver_init_node_id(
 
     bad_reinit = run_pageserver(good_init_cmd)
     assert bad_reinit.returncode == 1, "pageserver refuses to init if already exists"
-    assert "already exists, cannot init it" in bad_reinit.stderr
+    assert "config file already exists" in bad_reinit.stderr
-
-    bad_update = run_pageserver(["--update-config", "-c", "id = 3"])
-    assert bad_update.returncode == 1, "pageserver should not allow updating node id"
-    assert "has node id already, it cannot be overridden" in bad_update.stderr
 
 
 def check_client(env: NeonEnv, client: PageserverHttpClient):

test_runner/regress/test_pageserver_generations.py

@@ -220,7 +220,12 @@ def test_generations_upgrade(neon_env_builder: NeonEnvBuilder):
     # We will start a pageserver with no control_plane_api set, so it won't be able to self-register
     env.storage_controller.node_register(env.pageserver)
 
-    env.pageserver.start(overrides=('--pageserver-config-override=control_plane_api=""',))
+    replaced_config = env.pageserver.patch_config_toml_nonrecursive(
+        {
+            "control_plane_api": "",
+        }
+    )
+    env.pageserver.start()
     env.storage_controller.node_configure(env.pageserver.id, {"availability": "Active"})
 
     env.neon_cli.create_tenant(
@@ -251,8 +256,8 @@ def test_generations_upgrade(neon_env_builder: NeonEnvBuilder):
         assert parse_generation_suffix(key) is None
 
     env.pageserver.stop()
-
     # Starting without the override that disabled control_plane_api
+    env.pageserver.patch_config_toml_nonrecursive(replaced_config)
     env.pageserver.start()
 
     generate_uploads_and_deletions(env, pageserver=env.pageserver, init=False)
@@ -525,9 +530,12 @@ def test_emergency_mode(neon_env_builder: NeonEnvBuilder, pg_bin: PgBin):
     # incident, but it might be unavoidable: if so, we want to be able to start up
     # and serve clients.
     env.pageserver.stop()  # Non-immediate: implicitly checking that shutdown doesn't hang waiting for CP
-    env.pageserver.start(
+    replaced = env.pageserver.patch_config_toml_nonrecursive(
-        overrides=("--pageserver-config-override=control_plane_emergency_mode=true",),
+        {
+            "control_plane_emergency_mode": True,
+        }
     )
+    env.pageserver.start()
 
     # The pageserver should provide service to clients
     generate_uploads_and_deletions(env, init=False, pageserver=env.pageserver)
@@ -549,6 +557,7 @@ def test_emergency_mode(neon_env_builder: NeonEnvBuilder, pg_bin: PgBin):
 
     # The pageserver should work fine when subsequently restarted in non-emergency mode
     env.pageserver.stop()  # Non-immediate: implicitly checking that shutdown doesn't hang waiting for CP
+    env.pageserver.patch_config_toml_nonrecursive(replaced)
     env.pageserver.start()
 
     generate_uploads_and_deletions(env, init=False, pageserver=env.pageserver)

test_runner/regress/test_remote_storage.py

@@ -1,6 +1,3 @@
-# It's possible to run any regular test with the local fs remote storage via
-# env NEON_PAGESERVER_OVERRIDES="remote_storage={local_path='/tmp/neon_zzz/'}" poetry ......
-
 import os
 import queue
 import shutil

test_runner/regress/test_storage_controller.py

@@ -290,9 +290,12 @@ def test_storage_controller_onboarding(neon_env_builder: NeonEnvBuilder, warm_up
     # This is the pageserver where we'll initially create the tenant.  Run it in emergency
     # mode so that it doesn't talk to storage controller, and do not register it.
     env.pageservers[0].allowed_errors.append(".*Emergency mode!.*")
-    env.pageservers[0].start(
+    env.pageservers[0].patch_config_toml_nonrecursive(
-        overrides=("--pageserver-config-override=control_plane_emergency_mode=true",),
+        {
+            "control_plane_emergency_mode": True,
+        }
     )
+    env.pageservers[0].start()
     origin_ps = env.pageservers[0]
 
     # These are the pageservers managed by the sharding service, where the tenant