mirrors/scylladb
1
0
Fork 0
mirror of https://github.com/scylladb/scylladb.git synced 2026-04-27 20:05:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

sstables: initial work on handling a partially written sstable

Browse Source 
The solution was proposed by Nadav. When writing a new sstable,
write all usual files, write the TOC to a temporary file, and
then rename it, which is atomic.
Files not belonging to any TOC are invalid, so we ensure that
partially written sstables aren't reused.

Avi also proposed using fsync on the sstable directory to guarantee
that the files reached the disk before sealing the sstable.

Subsequently, we should add code to avoid loading sstable which
TOC is either temporary or doesn't exist. Temporary TOC files
should also be deleted.

Signed-off-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
This commit is contained in:
Raphael S. Carvalho
2015-08-13 12:37:24 -03:00
committed by Avi Kivity
parent cb55bdbfbc
commit 82425fd24a
2 changed files with 25 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
sstables/sstables.cc
View File

@@ -679,7 +679,9 @@ future<> sstable::read_toc() {
}
void sstable::write_toc() {
auto file_path = filename(sstable::component_type::TOC);
// Create TOC file with the string 'tmp-' prepended to it, meaning TOC
// is a temporary file.
auto file_path = temporary_filename(sstable::component_type::TOC);
sstlog.debug("Writing TOC file {} ", file_path);
@@ -695,6 +697,16 @@ void sstable::write_toc() {
}
w.flush().get();
w.close().get();
file dir_f = engine().open_directory(_dir).get0();
// Guarantee that every component of this sstable reached the disk.
dir_f.flush().get();
// Rename TOC because it's no longer temporary.
engine().rename_file(file_path, filename(sstable::component_type::TOC)).get();
// Guarantee that the changes above reached the disk.
dir_f.flush().get();
dir_f.close().get();
// If this point was reached, sstable should be safe in disk.
}
void write_crc(const sstring file_path, checksum& c) {
@@ -1368,8 +1380,12 @@ const sstring sstable::filename(component_type f) {
return filename(_dir, _ks, _cf, _version, _generation, _format, f);
}
const sstring sstable::temporary_filename(component_type f) {
return filename(_dir, _ks, _cf, _version, _generation, _format, f, true);
}
const sstring sstable::filename(sstring dir, sstring ks, sstring cf, version_types version, unsigned long generation,
format_types format, component_type component) {
format_types format, component_type component, bool temporary) {
static std::unordered_map<version_types, std::function<sstring (entry_descriptor d)>, enum_hash<version_types>> strmap = {
{ sstable::version_types::ka, [] (entry_descriptor d) {
@@ -1380,7 +1396,11 @@ const sstring sstable::filename(sstring dir, sstring ks, sstring cf, version_typ
}
};
return dir + "/" + strmap[version](entry_descriptor(ks, cf, version, generation, format, component));
if (temporary) {
return dir + "/tmp-" + strmap[version](entry_descriptor(ks, cf, version, generation, format, component));
} else {
return dir + "/" + strmap[version](entry_descriptor(ks, cf, version, generation, format, component));
}
}
entry_descriptor entry_descriptor::make_descriptor(sstring fname) {

3
sstables/sstables.hh
View File

@@ -167,7 +167,7 @@ public:
static version_types version_from_sstring(sstring& s);
static format_types format_from_sstring(sstring& s);
static const sstring filename(sstring dir, sstring ks, sstring cf, version_types version, unsigned long generation,
format_types format, component_type component);
format_types format, component_type component, bool temporary = false);
future<> load();
@@ -281,6 +281,7 @@ private:
const bool has_component(component_type f);
const sstring filename(component_type f);
const sstring temporary_filename(component_type f);
template <sstable::component_type Type, typename T>
future<> read_simple(T& comp);