ahuston-0/hydra
1
0
Fork 0
Code Issues Pull Requests 1 Packages Projects Releases Wiki Activity

Automatically retry aborted builds

Browse Source 
Aborted builds are now put back on the runnable queue and retried
after a certain time interval (currently 60 seconds for the first
retry, then tripled on each subsequent retry).
This commit is contained in:
Eelco Dolstra
2015-06-17 11:45:20 +02:00
parent e02654b3a0
commit b91a616520
4 changed files with 145 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
src/hydra-queue-runner/build-remote.cc
View File

@@ -84,7 +84,7 @@ static void copyClosureTo(std::shared_ptr<StoreAPI> store,
for (auto i = sorted.rbegin(); i != sorted.rend(); ++i)
if (present.find(*i) == present.end()) missing.push_back(*i);
printMsg(lvlError, format("sending %1% missing paths") % missing.size());
printMsg(lvlDebug, format("sending %1% missing paths") % missing.size());
writeInt(cmdImportPaths, to);
exportPaths(*store, missing, false, to);
@@ -128,23 +128,28 @@ void buildRemote(std::shared_ptr<StoreAPI> store,
FdSink to(child.to);
/* Handshake. */
writeInt(SERVE_MAGIC_1, to);
writeInt(SERVE_PROTOCOL_VERSION, to);
to.flush();
try {
writeInt(SERVE_MAGIC_1, to);
writeInt(SERVE_PROTOCOL_VERSION, to);
to.flush();
unsigned int magic = readInt(from);
if (magic != SERVE_MAGIC_2)
throw Error(format("protocol mismatch with ‘nix-store --serve’ on ‘%1%’") % sshName);
unsigned int version = readInt(from);
if (GET_PROTOCOL_MAJOR(version) != 0x200)
throw Error(format("unsupported ‘nix-store --serve’ protocol version on ‘%1%’") % sshName);
unsigned int magic = readInt(from);
if (magic != SERVE_MAGIC_2)
throw Error(format("protocol mismatch with ‘nix-store --serve’ on ‘%1%’") % sshName);
unsigned int version = readInt(from);
if (GET_PROTOCOL_MAJOR(version) != 0x200)
throw Error(format("unsupported ‘nix-store --serve’ protocol version on ‘%1%’") % sshName);
} catch (EndOfFile & e) {
child.pid.wait(true);
throw Error(format("cannot connect to ‘%1%’: %2%") % sshName % chomp(readFile(logFile)));
}
/* Copy the input closure. */
printMsg(lvlError, format("sending closure of ‘%1%’ to ‘%2%’") % drvPath % sshName);
printMsg(lvlDebug, format("sending closure of ‘%1%’ to ‘%2%’") % drvPath % sshName);
copyClosureTo(store, from, to, PathSet({drvPath}));
/* Do the build. */
printMsg(lvlError, format("building ‘%1%’ on ‘%2%’") % drvPath % sshName);
printMsg(lvlDebug, format("building ‘%1%’ on ‘%2%’") % drvPath % sshName);
writeInt(cmdBuildPaths, to);
writeStrings(PathSet({drvPath}), to);
writeInt(3600, to); // == maxSilentTime, FIXME
@@ -162,7 +167,7 @@ void buildRemote(std::shared_ptr<StoreAPI> store,
}
/* Copy the output paths. */
printMsg(lvlError, format("copying outputs of ‘%1%’ from ‘%2%’") % drvPath % sshName);
printMsg(lvlDebug, format("copying outputs of ‘%1%’ from ‘%2%’") % drvPath % sshName);
PathSet outputs;
for (auto & output : drv.outputs)
outputs.insert(output.second.path);