maurice/snix
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

* Use polling to wait for a remote build slot when using a build hook

Browse source 
(that is, call the build hook with a certain interval until it
  accepts the build).
* build-remote.pl was totally broken: for all system types other than
  the local system type, it would send all builds to the *first*
  machine of the appropriate type.
This commit is contained in:
Eelco Dolstra 2009-03-29 18:00:45 +00:00
parent 47706e3924
commit 737423a89c
2 changed files with 3 additions and 36 deletions
Download patch file Download diff file Expand all files Collapse all files

7
scripts/build-remote.pl.in
View file

@ -23,8 +23,7 @@ use English '-no_match_vars';
my $loadIncreased = 0; my $loadIncreased = 0;
my ($amWilling, $localSystem, $neededSystem, $drvPath, $mustRun, $maxSilentTime) = @ARGV; my ($amWilling, $localSystem, $neededSystem, $drvPath, $maxSilentTime) = @ARGV;
$mustRun = 0 unless defined $mustRun;
$maxSilentTime = 0 unless defined $maxSilentTime; $maxSilentTime = 0 unless defined $maxSilentTime;
sub sendReply { sub sendReply {
@ -87,12 +86,10 @@ LOOP: foreach my $cur (@machines) {
# We have a machine of the right type. Try to get a lock on # We have a machine of the right type. Try to get a lock on
# one of the machine's lock files. # one of the machine's lock files.
my $slot = 0; my $slot = 0;
while ($slot < $cur->{maxJobs} || ($mustRun && !$canBuildLocally)) { while ($slot < $cur->{maxJobs}) {
my $slotLock = "$currentLoad/" . $cur->{systemType} . "-" . $cur->{hostName} . "-$slot"; my $slotLock = "$currentLoad/" . $cur->{systemType} . "-" . $cur->{hostName} . "-$slot";
open SLOTLOCK, ">>$slotLock" or die; open SLOTLOCK, ">>$slotLock" or die;
if (flock(SLOTLOCK, LOCK_EX | LOCK_NB)) { if (flock(SLOTLOCK, LOCK_EX | LOCK_NB)) {
print STDERR "warning: exceeding maximum load on " . $cur->{systemType} . "\n"
if $slot >= $cur->{maxJobs};
$machine = $cur; $machine = $cur;
last LOOP; last LOOP;
} }

32
src/libstore/build.cc
View file

@ -52,9 +52,6 @@ namespace nix {
using std::map; using std::map;
/* !!! TODO derivationFromPath shouldn't be used here */
static string pathNullDevice = "/dev/null"; static string pathNullDevice = "/dev/null";
@ -229,9 +226,6 @@ public:
/* Can we start another child process? */ /* Can we start another child process? */
bool canBuildMore(); bool canBuildMore();
/* Can we postpone a build right now? */
bool canPostpone();
/* Registers a running child process. `inBuildSlot' means that /* Registers a running child process. `inBuildSlot' means that
the process counts towards the jobs limit. */ the process counts towards the jobs limit. */
void childStarted(GoalPtr goal, pid_t pid, void childStarted(GoalPtr goal, pid_t pid,
@ -247,10 +241,6 @@ public:
might be right away). */ might be right away). */
void waitForBuildSlot(GoalPtr goal); void waitForBuildSlot(GoalPtr goal);
/* Put `goal' to sleep until a child process terminates, i.e., a
call is made to childTerminate(..., true). */
void waitForChildTermination(GoalPtr goal);
/* Wait for any goal to finish. Pretty indiscriminate way to /* Wait for any goal to finish. Pretty indiscriminate way to
wait for some resource that some other goal is holding. */ wait for some resource that some other goal is holding. */
void waitForAnyGoal(GoalPtr goal); void waitForAnyGoal(GoalPtr goal);
@ -1041,7 +1031,7 @@ void DerivationGoal::tryToBuild()
return; return;
case rpPostpone: case rpPostpone:
/* Not now; wait until at least one child finishes. */ /* Not now; wait until at least one child finishes. */
worker.waitForChildTermination(shared_from_this()); worker.waitForAWhile(shared_from_this());
outputLocks.unlock(); outputLocks.unlock();
return; return;
case rpDecline: case rpDecline:
@ -1246,7 +1236,6 @@ DerivationGoal::HookReply DerivationGoal::tryBuildHook()
thisSystem.c_str(), thisSystem.c_str(),
drv.platform.c_str(), drv.platform.c_str(),
drvPath.c_str(), drvPath.c_str(),
(worker.canPostpone() ? (string) "0" : "1").c_str(),
(format("%1%") % maxSilentTime).str().c_str(), (format("%1%") % maxSilentTime).str().c_str(),
NULL); NULL);
@ -2243,8 +2232,6 @@ void SubstitutionGoal::tryToRun()
logPipe.readSide.close(); logPipe.readSide.close();
/* !!! close other handles */
commonChildInit(logPipe); commonChildInit(logPipe);
/* Fill in the arguments. */ /* Fill in the arguments. */
@ -2286,7 +2273,6 @@ void SubstitutionGoal::finished()
/* Since we got an EOF on the logger pipe, the substitute is /* Since we got an EOF on the logger pipe, the substitute is
presumed to have terminated. */ presumed to have terminated. */
/* !!! this could block! */
pid_t savedPid = pid; pid_t savedPid = pid;
int status = pid.wait(true); int status = pid.wait(true);
@ -2469,12 +2455,6 @@ bool Worker::canBuildMore()
} }
bool Worker::canPostpone()
{
return children.size() != 0;
}
void Worker::childStarted(GoalPtr goal, void Worker::childStarted(GoalPtr goal,
pid_t pid, const set<int> & fds, bool inBuildSlot) pid_t pid, const set<int> & fds, bool inBuildSlot)
{ {
@ -2527,16 +2507,6 @@ void Worker::waitForBuildSlot(GoalPtr goal)
} }
void Worker::waitForChildTermination(GoalPtr goal)
{
debug("wait for child termination");
if (children.size() == 0)
throw Error("waiting for a build slot, yet there are no running children - "
"maybe the build hook gave an inappropriate `postpone' reply?");
wantingToBuild.insert(goal);
}
void Worker::waitForAnyGoal(GoalPtr goal) void Worker::waitForAnyGoal(GoalPtr goal)
{ {
debug("wait for any goal"); debug("wait for any goal");