bahman/scylla
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

commitlog: cycle based on total size, not on mutation size

Browse Source 
We calculate two sizes during the allocation: "size", which is the
in-segment size of this mutation, and "s", which is that plus the
overhead. cycle() must be called with the latter, not the former, as
doing otherwise may lead to buffer overflows.

Signed-off-by: Glauber Costa <glauber@scylladb.com>
Message-Id: <ccf346d8d0ebb44a1ba9fd069653bab0d7be0a61.1477063157.git.glauber@scylladb.com>
This commit is contained in:
Glauber Costa
2016-10-21 11:19:20 -04:00
committed by Avi Kivity
parent d9875784a1
commit a13c410749
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
db/commitlog/commitlog.cc
View File

@@ -734,7 +734,7 @@ public:
return finish_and_get_new().then([id, writer = std::move(writer), permit = std::move(permit)] (auto new_seg) mutable {
return new_seg->allocate(id, std::move(writer), std::move(permit));
});
} else if (!_buffer.empty() && (size > (_buffer.size() - _buf_pos))) { // enough data?
} else if (!_buffer.empty() && (s > (_buffer.size() - _buf_pos))) { // enough data?
if (_segment_manager->cfg.mode == sync_mode::BATCH) {
// TODO: this could cause starvation if we're really unlucky.
// If we run batch mode and find ourselves not fit in a non-empty