1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
|
// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include "maintenancecontroller.h"
#include "maintenancejobrunner.h"
#include "i_blockable_maintenance_job.h"
#include <vespa/searchcorespi/index/i_thread_service.h>
#include <vespa/searchcore/proton/common/scheduledexecutor.h>
#include <vespa/vespalib/util/lambdatask.h>
#include <vespa/vespalib/util/thread.h>
#include <vespa/log/log.h>
LOG_SETUP(".proton.server.maintenancecontroller");
using document::BucketId;
using vespalib::Executor;
using vespalib::MonitoredRefCount;
using vespalib::makeLambdaTask;
using vespalib::thread::as_zu;
namespace proton {
namespace {
class JobWrapperTask : public Executor::Task
{
private:
MaintenanceJobRunner *_job;
public:
JobWrapperTask(MaintenanceJobRunner *job) : _job(job) {}
void run() override { _job->run(); }
};
}
MaintenanceController::MaintenanceController(FNET_Transport & transport,
ISyncableThreadService& masterThread,
MonitoredRefCount& refCount,
const DocTypeName& docTypeName)
: _masterThread(masterThread),
_refCount(refCount),
_readySubDB(),
_remSubDB(),
_notReadySubDB(),
_periodicTimer(std::make_unique<ScheduledExecutor>(transport)),
_state(State::INITIALIZING),
_docTypeName(docTypeName),
_jobs(),
_jobsLock()
{ }
MaintenanceController::~MaintenanceController()
{
kill();
}
void
MaintenanceController::registerJob(IMaintenanceJob::UP job)
{
// Called by master write thread
Guard guard(_jobsLock);
_jobs.push_back(std::make_shared<MaintenanceJobRunner>(_masterThread, std::move(job)));
}
void
MaintenanceController::killJobs()
{
if (_state == State::STARTED) {
_state = State::PAUSED;
}
// Called by master write thread
assert(_masterThread.isCurrentThread());
LOG(debug, "killJobs(): threadId=%zu", as_zu(std::this_thread::get_id()));
_periodicTaskHandles.clear();
// No need to take _jobsLock as modification of _jobs also happens in master write thread.
for (auto &job : _jobs) {
job->stop(); // Make sure no more tasks are added to the executor
}
JobList tmpJobs;
{
Guard guard(_jobsLock);
tmpJobs.swap(_jobs);
}
// Hold jobs until existing tasks have been drained
_masterThread.execute(makeLambdaTask([this, jobs=std::move(tmpJobs)]() {
performHoldJobs(std::move(jobs));
}));
}
void
MaintenanceController::updateMetrics(DocumentDBTaggedMetrics & metrics)
{
Guard guard(_jobsLock);
for (auto &job : _jobs) {
job->getJob().updateMetrics(metrics); // Make sure no more tasks are added to the executor
}
}
void
MaintenanceController::performHoldJobs(JobList jobs)
{
// Called by master write thread
LOG(debug, "performHoldJobs(): threadId=%zu", as_zu(std::this_thread::get_id()));
(void) jobs;
}
void
MaintenanceController::stop()
{
assert(!_masterThread.isCurrentThread());
_masterThread.execute(makeLambdaTask([this]() { _state = State::STOPPING; killJobs(); }));
_masterThread.sync(); // Wait for killJobs()
_masterThread.sync(); // Wait for already scheduled maintenance jobs and performHoldJobs
}
searchcorespi::index::IThreadService &
MaintenanceController::masterThread() {
return _masterThread;
}
void
MaintenanceController::kill()
{
stop();
_readySubDB.clear();
_remSubDB.clear();
_notReadySubDB.clear();
}
void
MaintenanceController::start()
{
// Called by master write thread
assert(_state == State::INITIALIZING);
_state = State::STARTED;
restart();
}
void
MaintenanceController::restart()
{
// Called by master write thread
if (!getStarted() || getStopping() || !_readySubDB.valid()) {
return;
}
addJobsToPeriodicTimer();
}
void
MaintenanceController::addJobsToPeriodicTimer()
{
_periodicTaskHandles.clear();
// No need to take _jobsLock as modification of _jobs also happens in master write thread.
for (const auto &jw : _jobs) {
const IMaintenanceJob &job = jw->getJob();
LOG(debug, "addJobsToPeriodicTimer(): docType='%s', job.name='%s', job.delay=%f, job.interval=%f",
_docTypeName.getName().c_str(), job.getName().c_str(), vespalib::to_s(job.getDelay()),
vespalib::to_s(job.getInterval()));
if (job.getInterval() == vespalib::duration::zero()) {
jw->run();
continue;
}
_periodicTaskHandles.push_back(_periodicTimer->scheduleAtFixedRate(std::make_unique<JobWrapperTask>(jw.get()),
job.getDelay(), job.getInterval()));
}
}
void
MaintenanceController::newConfig()
{
// Called by master write thread
restart();
}
namespace {
void
assert_equal_meta_store_instances(const MaintenanceDocumentSubDB& old_db,
const MaintenanceDocumentSubDB& new_db)
{
if (old_db.valid() && new_db.valid()) {
assert(old_db.meta_store().get() == new_db.meta_store().get());
}
}
}
void
MaintenanceController::syncSubDBs(const MaintenanceDocumentSubDB &readySubDB,
const MaintenanceDocumentSubDB &remSubDB,
const MaintenanceDocumentSubDB ¬ReadySubDB)
{
// Called by master write thread
bool oldValid = _readySubDB.valid();
assert(readySubDB.valid());
assert(remSubDB.valid());
// Document meta store instances should not change. Maintenance jobs depend on this fact.
assert_equal_meta_store_instances(_readySubDB, readySubDB);
assert_equal_meta_store_instances(_remSubDB, remSubDB);
assert_equal_meta_store_instances(_notReadySubDB, notReadySubDB);
_readySubDB = readySubDB;
_remSubDB = remSubDB;
_notReadySubDB = notReadySubDB;
if (!oldValid && getStarted()) {
restart();
}
}
} // namespace proton
|
