Compare commits

...

41 Commits

Author SHA1 Message Date
John Ericson
08c9225a76 Merge branch 'nix-2.20' into hydra.nixos.org 2024-05-21 18:08:58 -04:00
John Ericson
21044bc4d9 flake.lock: Update
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/8f42912c80c0a03f62f6a3d28a3af05a9762565d' (2024-01-30)
  → 'github:NixOS/nix/ab48ea416a203e9ccefb70aa634e27477e4c1ac4' (2024-05-15)
2024-05-21 18:01:30 -04:00
John Ericson
b676b08fac Merge pull request #1368 from Ma27/login-submit-btn
Use `submit` event in login form
2024-03-26 11:23:51 -04:00
John Ericson
d614163e9c Merge pull request #1370 from Ma27/reconnect-db
hydra-queue-runner: drop broken connections from pool
2024-03-26 11:21:35 -04:00
Maximilian Bosch
99afff03b0 hydra-queue-runner: drop broken connections from pool
Closes #1336

When restarting postgresql, the connections are still reused in
`hydra-queue-runner` causing errors like this

    main thread: Lost connection to the database server.
    queue monitor: Lost connection to the database server.

and no more builds being processed.

`hydra-evaluator` doesn't have that issue since it crashes right away.
We could let it retry indefinitely as well (see below), but I don't
want to change too much.

If the DB is still unreachable 10s later, the process will stop with a
non-zero exit code because of a missing DB connection. This however
isn't such a big deal because it will be immediately restarted
afterwards. With the current configuration, Hydra will never give up,
but restart (and retry) infinitely. To me that seems reasonable, i.e. to
retry DB connections on a long-running process. If this doesn't work
out, the monitoring should fire anyways because the queue fills up, but
I'm open to discuss that.

Please note that this isn't reproducible with the DB and the queue
runner on the same machine when using `services.hydra-dev`, because of
the `Requires=` dependency `hydra-queue-runner.service` ->
`hydra-init.service` -> `postgresql.service` that causes the queue
runner to be restarted on `systemctl restart postgresql`.

Internally, Hydra uses Nix's pool data structure: it basically has N
slots (here DB connections) and whenever a new one is requested, an idle
slot is provided or a new one is created (when N slots are active, it'll
be waited until one slot is free). The issue in the code here is however
that whenever an error is encountered, the slot is released, however the
same broken connection will be reused the next time. By using
`Pool::Handle::markBad`, Nix will drop a broken slot. This is now being
done when `pqxx::broken_connection` was caught.
2024-03-15 14:09:31 +01:00
ajs124
8f56209bd6 Merge pull request #1361 from Ma27/fix-gitea-test
flake: fix gitea integration test
2024-03-08 15:28:07 +01:00
Maximilian Bosch
806c375c33 Don't send gitea status update when build is started
This was the source of a flaky test because sometimes hydra-notify was
quick enough to send out `buildStarted` and sometimes it apparently
wasn't which was quickly spottable with `nix build --rebuild`.

Removing that status update doesn't make a difference functionally,
gitea doesn't differentiate between "queued" and "running", so we send
the same status ("pending") out on both events, so we'd even safe one
avoidable request.
2024-03-08 11:07:38 +01:00
Maximilian Bosch
669617ab54 Use submit event in login form
It's a pet peeve from me when logging into my personal Hydra that I
always have to press the button rather than hitting Return after entering
my password.

Reason for that is that the form doesn't have a "submit" button, so far
it was always listened to the "click" event. Submit does that and you
can hit Return alternatively.
2024-03-07 18:49:52 +01:00
Janne Heß
c45c06509a Merge pull request #1364 from K900/urlencode-logs
urlencode drv names when fetching logs
2024-03-01 10:11:08 +01:00
K900
9db5d0a88d urlencode drv names when fetching logs
Otherwise names with special characters like + break things.
2024-02-26 22:48:16 +03:00
Maximilian Bosch
ceff5c5cfe flake: fix gitea integration test
This is an integration test that confirms that jobset definitions from
git repositories are correctly built and status updates pushed to the
gitea instance. The following things needed to be fixed:

* We're still on 23.05 where gitea is marked as insecure. Not going to
  update nixpkgs right now, but going for the quick fix.
* Since gitea 1.19 tokens have scopes that describe what's possible.
  Not specifying the scope in the DB appears to imply that no
  permissions are granted.
* Apparently we have three status updates now (for three status hooks,
  queued/started/finished). No idea why that was broken before, but the
  behavior still looks correct.
2024-02-12 18:30:03 +01:00
John Ericson
c1bd50a80d Merge pull request #1354 from NixOS/nix-2.20
Nix 2.20
2024-01-30 14:07:49 -05:00
John Ericson
14aabc1cc9 Update to released Nix 2.20
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/8df68a213fc52a57b02a57005b0e06cc8de40ce3' (2024-01-25)
  → 'github:NixOS/nix/8f42912c80c0a03f62f6a3d28a3af05a9762565d' (2024-01-30)
2024-01-30 13:33:20 -05:00
John Ericson
7b826ec5ad Merge branch 'nix-next' into nix-2.20 2024-01-30 13:26:45 -05:00
John Ericson
7a53b866f6 Merge branch 'master' into nix-next
• Updated input 'nix' (merge):
    'github:NixOS/nix/212ba69e6f995992f8b4e4c0656d19c0156c8714'
    'github:NixOS/nix/2c4bb93ba5a97e7078896ebc36385ce172960e4e' (2024-01-25)
  → 'github:NixOS/nix/8df68a213fc52a57b02a57005b0e06cc8de40ce3' (2024-01-25)
2024-01-25 16:26:07 -05:00
John Ericson
449eb2d873 Use more nix::Machine fields
The upstream fields were made to match Hydra, so we can get rid of the
extra fields temporary added in
70e5469303.
2024-01-24 20:14:31 -05:00
John Ericson
2bdbf51d7d flake.lock: Update
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/b6aee9a93f6646bbffd919d362a5c75c37bb9caa' (2024-01-23)
  → 'github:NixOS/nix/212ba69e6f995992f8b4e4c0656d19c0156c8714' (2024-01-24)
2024-01-24 18:46:56 -05:00
John Ericson
9e7ac58042 Merge branch 'master' into nix-next 2024-01-24 18:36:03 -05:00
John Ericson
9a86da0e7b Merge branch 'master' into nix-next 2024-01-23 15:49:14 -05:00
John Ericson
20b0ad3ba2 Merge pull request #1339 from NixOS/use-nix-ssh
Use Nix's `SSHMaster`
2024-01-23 10:35:02 -05:00
John Ericson
7386caaecf Use Nix's SSHMaster 2024-01-23 10:24:02 -05:00
John Ericson
84c46b6b68 Update to newer Nix
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/74534829f23b668fb9b2f2a14ff6afa4d5e71d4a' (2024-01-22)
  → 'github:NixOS/nix/b6aee9a93f6646bbffd919d362a5c75c37bb9caa' (2024-01-23)
2024-01-23 10:21:48 -05:00
John Ericson
f1d9230f25 Merge remote-tracking branch 'upstream/master' into nix-next 2024-01-23 01:18:13 -05:00
John Ericson
34c51fcea9 Merge pull request #1165 from obsidiansystems/factor-out-proto
Begin factoring out the protocol code
2024-01-22 14:49:07 -05:00
John Ericson
4ac31c89df Use nix::serv_proto::BasicConnection in build_remote.cc
- Use the type itself

  This lays the foundation for being able to dedup the protocol code.

- Use `BasicConnection::handshake`, replacing ours.

- Use `BasicConnection::queryValidPaths`

- Use `BasicConnection::putBuildDerivationRequest`
2024-01-22 14:20:39 -05:00
John Ericson
db7aa01b8d Update to newer Nix master
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/b7e016ab2464ad2e7e2e856ad0f173157135aae0' (2023-12-10)
  → 'github:NixOS/nix/74534829f23b668fb9b2f2a14ff6afa4d5e71d4a' (2024-01-22)
• Updated input 'nixpkgs':
    'github:NixOS/nixpkgs/e9f06adb793d1cca5384907b3b8a4071d5d7cb19' (2023-12-03)
  → 'github:NixOS/nixpkgs/a1982c92d8980a0114372973cbdfe0a307f1bdea' (2024-01-12)
• Removed input 'nix/lowdown-src'
2024-01-22 14:14:59 -05:00
John Ericson
89cfe26533 Merge remote-tracking branch 'upstream/master' into nix-next 2024-01-22 13:01:40 -05:00
John Ericson
f3a760ad9c Merge pull request #1324 from obsidiansystems/serve-proto-build-options-serializer
Use `ServeProto::Serialise<ServeProto::BuildOptions>`
2023-12-11 10:45:33 -05:00
John Ericson
8c10331ee8 Fix totalNarSize summation
I accidentally removed it in d0d3b0a298.
2023-12-10 14:05:26 -05:00
John Ericson
20f5a2120c Use ServeProto::Serialise<ServeProto::BuildOptions> 2023-12-10 13:24:17 -05:00
John Ericson
b56d2383c1 Do not attempt to speak a newer version of the protocol
Both sides need to agree on a version (with `std::min`) for anything to
work. Somehow... we've never done this.

With this comment, the next commit succeeds. Without this commit, the
next commit fails. This is because the next commit exposes serializers
which do different things for proto version 2.7, and we're currently
requesting 2.6.

Opened https://github.com/NixOS/nix/issues/9584 to track this issue
2023-12-10 13:24:17 -05:00
John Ericson
2bd67562b5 Merge pull request #1323 from obsidiansystems/serve-proto-build-options
Use `ServeProto::BuildOption`
2023-12-10 13:23:43 -05:00
John Ericson
69a5b00e60 Use ServeProto::BuildOption
More deduplication with Nix.
2023-12-10 13:01:00 -05:00
John Ericson
1d80b72ffb flake.lock: Update Nix master
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/c8458bd731eb1c74159bebe459ea00165e056b65' (2023-12-09)
  → 'github:NixOS/nix/b7e016ab2464ad2e7e2e856ad0f173157135aae0' (2023-12-10)
2023-12-10 13:00:47 -05:00
John Ericson
105fd18fee Merge pull request #1322 from NixOS/use-UnkeyedValidPathInfo-serializer
Use `ServeProto::Serialise<UnkeyedValidPathInfo>` for `QueryValidPaths`
2023-12-10 11:52:56 -05:00
John Ericson
f6f817926a std::move the into the path info map 2023-12-09 12:12:00 -05:00
John Ericson
d0d3b0a298 Use ServeProto::Serialise<UnkeyedValidPathInfo> for QueryValidPaths
Companion to already-merged https://github.com/NixOS/nix/pull/9560
2023-12-09 12:08:04 -05:00
John Ericson
3f932a6731 build-remote: Use std::map<StorePath, UnkeyedValidPathInfo>
It is less denormalized
2023-12-09 11:59:09 -05:00
John Ericson
aaa0e128c1 flake.lock: Update Nix master
Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/c3827ff6348a4d5199eaddf8dbc2ca2e2ef46ec5' (2023-12-07)
  → 'github:NixOS/nix/c8458bd731eb1c74159bebe459ea00165e056b65' (2023-12-09)
2023-12-09 11:59:02 -05:00
John Ericson
4515b5aa17 Merge pull request #1321 from NixOS/master
Mere `master` into `nix-next`
2023-12-09 11:53:58 -05:00
John Ericson
20c8263e3c Update to Nix master
The point of this branch is to always track Nix master, so we are
proactively ready to upgrade to the next Nix release when it is ready.

Flake lock file updates:

• Updated input 'nix':
    'github:NixOS/nix/50f8f1c8bc019a4c0fd098b9ac674b94cfc6af0d' (2023-11-27)
  → 'github:NixOS/nix/c3827ff6348a4d5199eaddf8dbc2ca2e2ef46ec5' (2023-12-07)
• Added input 'nix/libgit2':
    'github:libgit2/libgit2/45fd9ed7ae1a9b74b957ef4f337bc3c8b3df01b5' (2023-10-18)
2023-12-07 13:11:31 -05:00
11 changed files with 166 additions and 260 deletions

34
flake.lock generated
View File

@@ -16,58 +16,58 @@
"type": "github"
}
},
"lowdown-src": {
"libgit2": {
"flake": false,
"locked": {
"lastModified": 1633514407,
"narHash": "sha256-Dw32tiMjdK9t3ETl5fzGrutQTzh2rufgZV4A/BbxuD4=",
"owner": "kristapsdz",
"repo": "lowdown",
"rev": "d2c2b44ff6c27b936ec27358a2653caaef8f73b8",
"lastModified": 1697646580,
"narHash": "sha256-oX4Z3S9WtJlwvj0uH9HlYcWv+x1hqp8mhXl7HsLu2f0=",
"owner": "libgit2",
"repo": "libgit2",
"rev": "45fd9ed7ae1a9b74b957ef4f337bc3c8b3df01b5",
"type": "github"
},
"original": {
"owner": "kristapsdz",
"repo": "lowdown",
"owner": "libgit2",
"repo": "libgit2",
"type": "github"
}
},
"nix": {
"inputs": {
"flake-compat": "flake-compat",
"lowdown-src": "lowdown-src",
"libgit2": "libgit2",
"nixpkgs": [
"nixpkgs"
],
"nixpkgs-regression": "nixpkgs-regression"
},
"locked": {
"lastModified": 1706208340,
"narHash": "sha256-wNyHUEIiKKVs6UXrUzhP7RSJQv0A8jckgcuylzftl8k=",
"lastModified": 1715805674,
"narHash": "sha256-0CIBMECsA3ISJZrJcOTzi6wa3QENTKGLtOpYIOoxwxo=",
"owner": "NixOS",
"repo": "nix",
"rev": "2c4bb93ba5a97e7078896ebc36385ce172960e4e",
"rev": "ab48ea416a203e9ccefb70aa634e27477e4c1ac4",
"type": "github"
},
"original": {
"owner": "NixOS",
"ref": "2.19-maintenance",
"ref": "2.20-maintenance",
"repo": "nix",
"type": "github"
}
},
"nixpkgs": {
"locked": {
"lastModified": 1701615100,
"narHash": "sha256-7VI84NGBvlCTduw2aHLVB62NvCiZUlALLqBe5v684Aw=",
"lastModified": 1705033721,
"narHash": "sha256-K5eJHmL1/kev6WuqyqqbS1cdNnSidIZ3jeqJ7GbrYnQ=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "e9f06adb793d1cca5384907b3b8a4071d5d7cb19",
"rev": "a1982c92d8980a0114372973cbdfe0a307f1bdea",
"type": "github"
},
"original": {
"owner": "NixOS",
"ref": "nixos-23.05",
"ref": "nixos-23.05-small",
"repo": "nixpkgs",
"type": "github"
}

View File

@@ -1,8 +1,8 @@
{
description = "A Nix-based continuous build system";
inputs.nixpkgs.url = "github:NixOS/nixpkgs/nixos-23.05";
inputs.nix.url = "github:NixOS/nix/2.19-maintenance";
inputs.nixpkgs.url = "github:NixOS/nixpkgs/nixos-23.05-small";
inputs.nix.url = "github:NixOS/nix/2.20-maintenance";
inputs.nix.inputs.nixpkgs.follows = "nixpkgs";
# TODO get rid of this once https://github.com/NixOS/nix/pull/9546 is
@@ -180,13 +180,9 @@
root=d7f16a3412e01a43a414535b16007c6931d3a9c7
</gitea_authorization>
'';
nixpkgs.config.permittedInsecurePackages = [ "gitea-1.19.4" ];
nix = {
distributedBuilds = true;
buildMachines = [{
hostName = "localhost";
systems = [ system ];
}];
binaryCaches = [ ];
settings.substituters = [ ];
};
services.gitea = {
enable = true;
@@ -202,7 +198,7 @@
testScript =
let
scripts.mktoken = pkgs.writeText "token.sql" ''
INSERT INTO access_token (id, uid, name, created_unix, updated_unix, token_hash, token_salt, token_last_eight) VALUES (1, 1, 'hydra', 1617107360, 1617107360, 'a930f319ca362d7b49a4040ac0af74521c3a3c3303a86f327b01994430672d33b6ec53e4ea774253208686c712495e12a486', 'XRjWE9YW0g', '31d3a9c7');
INSERT INTO access_token (id, uid, name, created_unix, updated_unix, token_hash, token_salt, token_last_eight, scope) VALUES (1, 1, 'hydra', 1617107360, 1617107360, 'a930f319ca362d7b49a4040ac0af74521c3a3c3303a86f327b01994430672d33b6ec53e4ea774253208686c712495e12a486', 'XRjWE9YW0g', '31d3a9c7', 'all');
'';
scripts.git-setup = pkgs.writeShellScript "setup.sh" ''
@@ -357,9 +353,9 @@
response = json.loads(data)
assert len(response) == 2, "Expected exactly two status updates for latest commit!"
assert response[0]['status'] == "success", "Expected latest status to be success!"
assert response[1]['status'] == "pending", "Expected first status to be pending!"
assert len(response) == 2, "Expected exactly three status updates for latest commit (queued, finished)!"
assert response[0]['status'] == "success", "Expected finished status to be success!"
assert response[1]['status'] == "pending", "Expected queued status to be pending!"
machine.shutdown()
'';

View File

@@ -89,7 +89,7 @@ struct MyArgs : MixEvalArgs, MixCommonArgs, RootArgs
static MyArgs myArgs;
static std::string queryMetaStrings(EvalState & state, DrvInfo & drv, const std::string & name, const std::string & subAttribute)
static std::string queryMetaStrings(EvalState & state, PackageInfo & drv, const std::string & name, const std::string & subAttribute)
{
Strings res;
std::function<void(Value & v)> rec;
@@ -181,7 +181,7 @@ static void worker(
// CA derivations do not have static output paths, so we
// have to defensively not query output paths in case we
// encounter one.
DrvInfo::Outputs outputs = drv->queryOutputs(
PackageInfo::Outputs outputs = drv->queryOutputs(
!experimentalFeatureSettings.isEnabled(Xp::CaDerivations));
if (drv->querySystem() == "unknown")

View File

@@ -8,6 +8,7 @@
#include "build-result.hh"
#include "path.hh"
#include "serve-protocol.hh"
#include "serve-protocol-impl.hh"
#include "state.hh"
#include "current-process.hh"
#include "processes.hh"
@@ -20,12 +21,6 @@
using namespace nix;
static void append(Strings & dst, const Strings & src)
{
dst.insert(dst.end(), src.begin(), src.end());
}
namespace nix::build_remote {
static Strings extraStoreArgs(std::string & machine)
@@ -48,63 +43,31 @@ static Strings extraStoreArgs(std::string & machine)
return result;
}
static void openConnection(::Machine::ptr machine, Path tmpDir, int stderrFD, SSHMaster::Connection & child)
static std::unique_ptr<SSHMaster::Connection> openConnection(
::Machine::ptr machine, SSHMaster & master)
{
std::string pgmName;
Pipe to, from;
to.create();
from.create();
Strings argv;
Strings command = {"nix-store", "--serve", "--write"};
if (machine->isLocalhost()) {
pgmName = "nix-store";
argv = {"nix-store", "--builders", "", "--serve", "--write"};
command.push_back("--builders");
command.push_back("");
} else {
pgmName = "ssh";
auto sshName = machine->sshName;
Strings extraArgs = extraStoreArgs(sshName);
argv = {"ssh", sshName};
if (machine->sshKey != "") append(argv, {"-i", machine->sshKey});
if (machine->sshPublicHostKey != "") {
Path fileName = tmpDir + "/host-key";
auto p = machine->sshName.find("@");
std::string host = p != std::string::npos ? std::string(machine->sshName, p + 1) : machine->sshName;
writeFile(fileName, host + " " + machine->sshPublicHostKey + "\n");
append(argv, {"-oUserKnownHostsFile=" + fileName});
}
append(argv,
{ "-x", "-a", "-oBatchMode=yes", "-oConnectTimeout=60", "-oTCPKeepAlive=yes"
, "--", "nix-store", "--serve", "--write" });
append(argv, extraArgs);
command.splice(command.end(), extraStoreArgs(machine->sshName));
}
child.sshPid = startProcess([&]() {
restoreProcessContext();
if (dup2(to.readSide.get(), STDIN_FILENO) == -1)
throw SysError("cannot dup input pipe to stdin");
if (dup2(from.writeSide.get(), STDOUT_FILENO) == -1)
throw SysError("cannot dup output pipe to stdout");
if (dup2(stderrFD, STDERR_FILENO) == -1)
throw SysError("cannot dup stderr");
execvp(argv.front().c_str(), (char * *) stringsToCharPtrs(argv).data()); // FIXME: remove cast
throw SysError("cannot start %s", pgmName);
auto ret = master.startCommand(std::move(command), {
"-a", "-oBatchMode=yes", "-oConnectTimeout=60", "-oTCPKeepAlive=yes"
});
to.readSide = -1;
from.writeSide = -1;
child.in = to.writeSide.release();
child.out = from.readSide.release();
// XXX: determine the actual max value we can use from /proc.
// FIXME: Should this be upstreamed into `startCommand` in Nix?
int pipesize = 1024 * 1024;
fcntl(child.in.get(), F_SETPIPE_SZ, &pipesize);
fcntl(child.out.get(), F_SETPIPE_SZ, &pipesize);
fcntl(ret->in.get(), F_SETPIPE_SZ, &pipesize);
fcntl(ret->out.get(), F_SETPIPE_SZ, &pipesize);
return ret;
}
@@ -122,13 +85,10 @@ static void copyClosureTo(
garbage-collect paths that are already there. Optionally, ask
the remote host to substitute missing paths. */
// FIXME: substitute output pollutes our build log
conn.to << ServeProto::Command::QueryValidPaths << 1 << useSubstitutes;
ServeProto::write(destStore, conn, closure);
conn.to.flush();
/* Get back the set of paths that are already valid on the remote
host. */
auto present = ServeProto::Serialise<StorePathSet>::read(destStore, conn);
auto present = conn.queryValidPaths(
destStore, true, closure, useSubstitutes);
if (present.size() == closure.size()) return;
@@ -153,7 +113,7 @@ static void copyClosureTo(
// FIXME: use Store::topoSortPaths().
static StorePaths reverseTopoSortPaths(const std::map<StorePath, ValidPathInfo> & paths)
static StorePaths reverseTopoSortPaths(const std::map<StorePath, UnkeyedValidPathInfo> & paths)
{
StorePaths sorted;
StorePathSet visited;
@@ -194,28 +154,6 @@ static std::pair<Path, AutoCloseFD> openLogFile(const std::string & logDir, cons
return {std::move(logFile), std::move(logFD)};
}
/**
* @param conn is not fully initialized; it is this functions job to set
* the `remoteVersion` field after the handshake is completed.
* Therefore, no `ServeProto::Serialize` functions can be used until
* that field is set.
*/
static void handshake(::Machine::Connection & conn, unsigned int repeats)
{
conn.to << SERVE_MAGIC_1 << 0x206;
conn.to.flush();
unsigned int magic = readInt(conn.from);
if (magic != SERVE_MAGIC_2)
throw Error("protocol mismatch with nix-store --serve on %1%", conn.machine->sshName);
conn.remoteVersion = readInt(conn.from);
// Now `conn` is initialized.
if (GET_PROTOCOL_MAJOR(conn.remoteVersion) != 0x200)
throw Error("unsupported nix-store --serve protocol version on %1%", conn.machine->sshName);
if (GET_PROTOCOL_MINOR(conn.remoteVersion) < 3 && repeats > 0)
throw Error("machine %1% does not support repeating a build; please upgrade it to Nix 1.12", conn.machine->sshName);
}
static BasicDerivation sendInputs(
State & state,
Step & step,
@@ -286,21 +224,11 @@ static BuildResult performBuild(
Store & localStore,
StorePath drvPath,
const BasicDerivation & drv,
const State::BuildOptions & options,
const ServeProto::BuildOptions & options,
counter & nrStepsBuilding
)
{
conn.to << ServeProto::Command::BuildDerivation << localStore.printStorePath(drvPath);
writeDerivation(conn.to, localStore, drv);
conn.to << options.maxSilentTime << options.buildTimeout;
if (GET_PROTOCOL_MINOR(conn.remoteVersion) >= 2)
conn.to << options.maxLogSize;
if (GET_PROTOCOL_MINOR(conn.remoteVersion) >= 3) {
conn.to
<< options.repeats // == build-repeat
<< options.enforceDeterminism;
}
conn.to.flush();
conn.putBuildDerivationRequest(localStore, drvPath, drv, options);
BuildResult result;
@@ -350,7 +278,7 @@ static BuildResult performBuild(
return result;
}
static std::map<StorePath, ValidPathInfo> queryPathInfos(
static std::map<StorePath, UnkeyedValidPathInfo> queryPathInfos(
::Machine::Connection & conn,
Store & localStore,
StorePathSet & outputs,
@@ -359,30 +287,18 @@ static std::map<StorePath, ValidPathInfo> queryPathInfos(
{
/* Get info about each output path. */
std::map<StorePath, ValidPathInfo> infos;
std::map<StorePath, UnkeyedValidPathInfo> infos;
conn.to << ServeProto::Command::QueryPathInfos;
ServeProto::write(localStore, conn, outputs);
conn.to.flush();
while (true) {
auto storePathS = readString(conn.from);
if (storePathS == "") break;
auto deriver = readString(conn.from); // deriver
auto references = ServeProto::Serialise<StorePathSet>::read(localStore, conn);
readLongLong(conn.from); // download size
auto narSize = readLongLong(conn.from);
auto narHash = Hash::parseAny(readString(conn.from), htSHA256);
auto ca = ContentAddress::parseOpt(readString(conn.from));
readStrings<StringSet>(conn.from); // sigs
ValidPathInfo info(localStore.parseStorePath(storePathS), narHash);
assert(outputs.count(info.path));
info.references = references;
info.narSize = narSize;
auto storePath = localStore.parseStorePath(storePathS);
auto info = ServeProto::Serialise<UnkeyedValidPathInfo>::read(localStore, conn);
totalNarSize += info.narSize;
info.narHash = narHash;
info.ca = ca;
if (deriver != "")
info.deriver = localStore.parseStorePath(deriver);
infos.insert_or_assign(info.path, info);
infos.insert_or_assign(std::move(storePath), std::move(info));
}
return infos;
@@ -423,14 +339,16 @@ static void copyPathsFromRemote(
NarMemberDatas & narMembers,
Store & localStore,
Store & destStore,
const std::map<StorePath, ValidPathInfo> & infos
const std::map<StorePath, UnkeyedValidPathInfo> & infos
)
{
auto pathsSorted = reverseTopoSortPaths(infos);
for (auto & path : pathsSorted) {
auto & info = infos.find(path)->second;
copyPathFromRemote(conn, narMembers, localStore, destStore, info);
copyPathFromRemote(
conn, narMembers, localStore, destStore,
ValidPathInfo { path, info });
}
}
@@ -508,7 +426,7 @@ private:
void State::buildRemote(ref<Store> destStore,
MachineReservation::ptr & reservation,
::Machine::ptr machine, Step::ptr step,
const BuildOptions & buildOptions,
const ServeProto::BuildOptions & buildOptions,
RemoteResult & result, std::shared_ptr<ActiveStep> activeStep,
std::function<void(StepState)> updateStep,
NarMemberDatas & narMembers)
@@ -519,21 +437,26 @@ void State::buildRemote(ref<Store> destStore,
AutoDelete logFileDel(logFile, false);
result.logFile = logFile;
nix::Path tmpDir = createTempDir();
AutoDelete tmpDirDel(tmpDir, true);
try {
updateStep(ssConnecting);
SSHMaster master {
machine->sshName,
machine->sshKey,
machine->sshPublicHostKey,
false, // no SSH master yet
false, // no compression yet
logFD.get(),
};
// FIXME: rewrite to use Store.
SSHMaster::Connection child;
build_remote::openConnection(machine, tmpDir, logFD.get(), child);
auto child = build_remote::openConnection(machine, master);
{
auto activeStepState(activeStep->state_.lock());
if (activeStepState->cancelled) throw Error("step cancelled");
activeStepState->pid = child.sshPid;
activeStepState->pid = child->sshPid;
}
Finally clearPid([&]() {
@@ -549,9 +472,13 @@ void State::buildRemote(ref<Store> destStore,
});
::Machine::Connection conn {
.from = child.out.get(),
.to = child.in.get(),
.machine = machine,
{
.to = child->in.get(),
.from = child->out.get(),
/* Handshake. */
.remoteVersion = 0xdadbeef, // FIXME avoid dummy initialize
},
/*.machine =*/ machine,
};
Finally updateStats([&]() {
@@ -559,14 +486,26 @@ void State::buildRemote(ref<Store> destStore,
bytesSent += conn.to.written;
});
constexpr ServeProto::Version our_version = 0x206;
try {
build_remote::handshake(conn, buildOptions.repeats);
conn.remoteVersion = decltype(conn)::handshake(
conn.to,
conn.from,
our_version,
machine->sshName);
} catch (EndOfFile & e) {
child.sshPid.wait();
child->sshPid.wait();
std::string s = chomp(readFile(result.logFile));
throw Error("cannot connect to %1%: %2%", machine->sshName, s);
}
// Do not attempt to speak a newer version of the protocol.
//
// Per https://github.com/NixOS/nix/issues/9584 should be handled as
// part of `handshake` in upstream nix.
conn.remoteVersion = std::min(conn.remoteVersion, our_version);
{
auto info(machine->state->connectInfo.lock());
info->consecutiveFailures = 0;
@@ -686,8 +625,8 @@ void State::buildRemote(ref<Store> destStore,
}
/* Shut down the connection. */
child.in = -1;
child.sshPid.wait();
child->in = -1;
child->sshPid.wait();
} catch (Error & e) {
/* Disable this machine until a certain period of time has

View File

@@ -100,10 +100,13 @@ State::StepResult State::doBuildStep(nix::ref<Store> destStore,
it). */
BuildID buildId;
std::optional<StorePath> buildDrvPath;
BuildOptions buildOptions;
buildOptions.repeats = step->isDeterministic ? 1 : 0;
buildOptions.maxLogSize = maxLogSize;
buildOptions.enforceDeterminism = step->isDeterministic;
// Other fields set below
nix::ServeProto::BuildOptions buildOptions {
.maxLogSize = maxLogSize,
.nrRepeats = step->isDeterministic ? 1u : 0u,
.enforceDeterminism = step->isDeterministic,
.keepFailed = false,
};
auto conn(dbPool.get());
@@ -138,7 +141,7 @@ State::StepResult State::doBuildStep(nix::ref<Store> destStore,
{
auto i = jobsetRepeats.find(std::make_pair(build2->projectName, build2->jobsetName));
if (i != jobsetRepeats.end())
buildOptions.repeats = std::max(buildOptions.repeats, i->second);
buildOptions.nrRepeats = std::max(buildOptions.nrRepeats, i->second);
}
}
if (!build) build = *dependents.begin();
@@ -149,7 +152,7 @@ State::StepResult State::doBuildStep(nix::ref<Store> destStore,
buildOptions.buildTimeout = build->buildTimeout;
printInfo("performing step %s %d times on %s (needed by build %d and %d others)",
localStore->printStorePath(step->drvPath), buildOptions.repeats + 1, machine->sshName, buildId, (dependents.size() - 1));
localStore->printStorePath(step->drvPath), buildOptions.nrRepeats + 1, machine->sshName, buildId, (dependents.size() - 1));
}
if (!buildOneDone)

View File

@@ -243,11 +243,11 @@ system_time State::doDispatch()
sort(machinesSorted.begin(), machinesSorted.end(),
[](const MachineInfo & a, const MachineInfo & b) -> bool
{
float ta = std::round(a.currentJobs / a.machine->speedFactorFloat);
float tb = std::round(b.currentJobs / b.machine->speedFactorFloat);
float ta = std::round(a.currentJobs / a.machine->speedFactor);
float tb = std::round(b.currentJobs / b.machine->speedFactor);
return
ta != tb ? ta < tb :
a.machine->speedFactorFloat != b.machine->speedFactorFloat ? a.machine->speedFactorFloat > b.machine->speedFactorFloat :
a.machine->speedFactor != b.machine->speedFactor ? a.machine->speedFactor > b.machine->speedFactor :
a.currentJobs > b.currentJobs;
});

View File

@@ -177,16 +177,16 @@ void State::parseMachines(const std::string & contents)
auto machine = std::make_shared<::Machine>(nix::Machine {
// `storeUri`, not yet used
"",
// `systemTypes`, not yet used
{},
// `systemTypes`
tokenizeString<StringSet>(tokens[1], ","),
// `sshKey`
tokens[2] == "-" ? "" : tokens[2],
// `maxJobs`
tokens[3] != ""
? string2Int<MaxJobs>(tokens[3]).value()
: 1,
// `speedFactor`, not yet used
1,
// `speedFactor`
atof(tokens[4].c_str()),
// `supportedFeatures`
std::move(supportedFeatures),
// `mandatoryFeatures`
@@ -198,8 +198,6 @@ void State::parseMachines(const std::string & contents)
});
machine->sshName = tokens[0];
machine->systemTypesSet = tokenizeString<StringSet>(tokens[1], ",");
machine->speedFactorFloat = atof(tokens[4].c_str());
/* Re-use the State object of the previous machine with the
same name. */
@@ -663,7 +661,7 @@ void State::dumpStatus(Connection & conn)
json machine = {
{"enabled", m->enabled},
{"systemTypes", m->systemTypesSet},
{"systemTypes", m->systemTypes},
{"supportedFeatures", m->supportedFeatures},
{"mandatoryFeatures", m->mandatoryFeatures},
{"nrStepsDone", s->nrStepsDone.load()},

View File

@@ -6,7 +6,46 @@
using namespace nix;
struct Extractor : ParseSink
struct NarMemberConstructor : CreateRegularFileSink
{
NarMemberData & curMember;
HashSink hashSink = HashSink { HashAlgorithm::SHA256 };
std::optional<uint64_t> expectedSize;
NarMemberConstructor(NarMemberData & curMember)
: curMember(curMember)
{ }
void isExecutable() override
{
}
void preallocateContents(uint64_t size) override
{
expectedSize = size;
}
void operator () (std::string_view data) override
{
assert(expectedSize);
*curMember.fileSize += data.size();
hashSink(data);
if (curMember.contents) {
curMember.contents->append(data);
}
assert(curMember.fileSize <= expectedSize);
if (curMember.fileSize == expectedSize) {
auto [hash, len] = hashSink.finish();
assert(curMember.fileSize == len);
curMember.sha256 = hash;
}
}
};
struct Extractor : FileSystemObjectSink
{
std::unordered_set<Path> filesToKeep {
"/nix-support/hydra-build-products",
@@ -15,7 +54,6 @@ struct Extractor : ParseSink
};
NarMemberDatas & members;
NarMemberData * curMember = nullptr;
Path prefix;
Extractor(NarMemberDatas & members, const Path & prefix)
@@ -27,53 +65,22 @@ struct Extractor : ParseSink
members.insert_or_assign(prefix + path, NarMemberData { .type = SourceAccessor::Type::tDirectory });
}
void createRegularFile(const Path & path) override
void createRegularFile(const Path & path, std::function<void(CreateRegularFileSink &)> func) override
{
curMember = &members.insert_or_assign(prefix + path, NarMemberData {
.type = SourceAccessor::Type::tRegular,
.fileSize = 0,
.contents = filesToKeep.count(path) ? std::optional("") : std::nullopt,
}).first->second;
}
std::optional<uint64_t> expectedSize;
std::unique_ptr<HashSink> hashSink;
void preallocateContents(uint64_t size) override
{
expectedSize = size;
hashSink = std::make_unique<HashSink>(htSHA256);
}
void receiveContents(std::string_view data) override
{
assert(expectedSize);
assert(curMember);
assert(hashSink);
*curMember->fileSize += data.size();
(*hashSink)(data);
if (curMember->contents) {
curMember->contents->append(data);
}
assert(curMember->fileSize <= expectedSize);
if (curMember->fileSize == expectedSize) {
auto [hash, len] = hashSink->finish();
assert(curMember->fileSize == len);
curMember->sha256 = hash;
hashSink.reset();
}
NarMemberConstructor nmc {
members.insert_or_assign(prefix + path, NarMemberData {
.type = SourceAccessor::Type::tRegular,
.fileSize = 0,
.contents = filesToKeep.count(path) ? std::optional("") : std::nullopt,
}).first->second,
};
func(nmc);
}
void createSymlink(const Path & path, const std::string & target) override
{
members.insert_or_assign(prefix + path, NarMemberData { .type = SourceAccessor::Type::tSymlink });
}
void isExecutable() override
{ }
void closeRegularFile() override
{ }
};

View File

@@ -726,7 +726,7 @@ BuildOutput State::getBuildOutputCached(Connection & conn, nix::ref<nix::Store>
product.fileSize = row[2].as<off_t>();
}
if (!row[3].is_null())
product.sha256hash = Hash::parseAny(row[3].as<std::string>(), htSHA256);
product.sha256hash = Hash::parseAny(row[3].as<std::string>(), HashAlgorithm::SHA256);
if (!row[4].is_null())
product.path = row[4].as<std::string>();
product.name = row[5].as<std::string>();

View File

@@ -23,6 +23,7 @@
#include "sync.hh"
#include "nar-extractor.hh"
#include "serve-protocol.hh"
#include "serve-protocol-impl.hh"
#include "machines.hh"
@@ -245,14 +246,6 @@ struct Machine : nix::Machine
we are not yet used to, but once we are, we don't need this. */
std::string sshName;
/* TODO Get rid once `nix::Machine::systemTypes` is a set not
vector. */
std::set<std::string> systemTypesSet;
/* TODO Get rid once `nix::Machine::systemTypes` is a `float` not
an `int`. */
float speedFactorFloat = 1.0;
struct State {
typedef std::shared_ptr<State> ptr;
counter currentJobs{0};
@@ -279,7 +272,7 @@ struct Machine : nix::Machine
{
/* Check that this machine is of the type required by the
step. */
if (!systemTypesSet.count(step->drv->platform == "builtin" ? nix::settings.thisSystem : step->drv->platform))
if (!systemTypes.count(step->drv->platform == "builtin" ? nix::settings.thisSystem : step->drv->platform))
return false;
/* Check that the step requires all mandatory features of this
@@ -309,29 +302,9 @@ struct Machine : nix::Machine
}
// A connection to a machine
struct Connection {
nix::FdSource from;
nix::FdSink to;
nix::ServeProto::Version remoteVersion;
struct Connection : nix::ServeProto::BasicClientConnection {
// Backpointer to the machine
ptr machine;
operator nix::ServeProto::ReadConn ()
{
return {
.from = from,
.version = remoteVersion,
};
}
operator nix::ServeProto::WriteConn ()
{
return {
.to = to,
.version = remoteVersion,
};
}
};
};
@@ -470,7 +443,7 @@ private:
/* How often the build steps of a jobset should be repeated in
order to detect non-determinism. */
std::map<std::pair<std::string, std::string>, unsigned int> jobsetRepeats;
std::map<std::pair<std::string, std::string>, size_t> jobsetRepeats;
bool uploadLogsToBinaryCache;
@@ -504,12 +477,6 @@ private:
public:
State(std::optional<std::string> metricsAddrOpt);
struct BuildOptions {
unsigned int maxSilentTime, buildTimeout, repeats;
size_t maxLogSize;
bool enforceDeterminism;
};
private:
nix::MaintainCount<counter> startDbUpdate();
@@ -600,7 +567,7 @@ private:
void buildRemote(nix::ref<nix::Store> destStore,
MachineReservation::ptr & reservation,
Machine::ptr machine, Step::ptr step,
const BuildOptions & buildOptions,
const nix::ServeProto::BuildOptions & buildOptions,
RemoteResult & result, std::shared_ptr<ActiveStep> activeStep,
std::function<void(StepState)> updateStep,
NarMemberDatas & narMembers);

View File

@@ -88,10 +88,6 @@ sub buildQueued {
common(@_, [], 0);
}
sub buildStarted {
common(@_, [], 1);
}
sub buildFinished {
common(@_, 2);
}