- 21 9月, 2017 1 次提交
-
-
由 Fabian Reinartz 提交于
IDs for new series are handed out before the postings are locked. Thus series are not indexed in order of their IDs, which could result in only partially sorted postings list. Iterating over those silently skipped elements as the sort invariant was violated.
-
- 19 9月, 2017 2 次提交
-
-
由 Goutham Veeramachaneni 提交于
Simplify series create logc in head
-
由 Fabian Reinartz 提交于
-
- 18 9月, 2017 5 次提交
-
-
由 Fabian Reinartz 提交于
-
由 Goutham Veeramachaneni 提交于
Add missing unlock on early return
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
- 13 9月, 2017 3 次提交
-
-
由 Fabian Reinartz 提交于
Expose NewIndexReader() and cleanups
-
由 Goutham Veeramachaneni 提交于
Signed-off-by: NGoutham Veeramachaneni <cs14btech11014@iith.ac.in>
-
由 Fabian Reinartz 提交于
replaced cobra with kingpin for the tsdb cli tool
-
- 11 9月, 2017 5 次提交
-
-
由 Fabian Reinartz 提交于
Fix various races
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
Fixup allocation regression during compaction
-
由 Fabian Reinartz 提交于
Compress the series chunk details in index.
-
- 10 9月, 2017 1 次提交
-
-
由 Krasi Georgiev 提交于
Signed-off-by: NKrasi Georgiev <krasi.root@gmail.com>
-
- 09 9月, 2017 3 次提交
-
-
由 Fabian Reinartz 提交于
When decoding data from mmaped blocks, we would like to retrieve a string backed by the mmaped region. As the underlying byte slice never changes, this is safe.
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
We were still fsyncing while holding the write lock when we cut a new segment. Given we cannot do anything but logging errors, we might just as well complete segments asynchronously. There's not realistic use case where one would fsync after every WAL entry, thus make the default of a flush interval of 0 to never fsync which is a much more likely use case.
-
- 08 9月, 2017 3 次提交
-
-
由 Goutham Veeramachaneni 提交于
Signed-off-by: NGoutham Veeramachaneni <cs14btech11014@iith.ac.in>
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
- 07 9月, 2017 2 次提交
-
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
- 06 9月, 2017 2 次提交
-
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
- 05 9月, 2017 1 次提交
-
-
由 Fabian Reinartz 提交于
This adds various new locks to replace the single big lock on the head. All parts now must be COW as they may be held by clients after initial retrieval. Series by ID and hashes are now held in a stripe lock to reduce contention and total holding time during GC. This should reduce starvation of readers.
-
- 04 9月, 2017 3 次提交
-
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Goutham Veeramachaneni 提交于
Signed-off-by: NGoutham Veeramachaneni <cs14btech11014@iith.ac.in>
-
- 01 9月, 2017 9 次提交
-
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
-
由 Fabian Reinartz 提交于
This changes the structure to a single WAL backed by a single head block. Parts of the head block can be compacted. This relieves us from any head amangement and greatly simplifies any consistency and isolation concerns by just having a single head.
-
由 Fabian Reinartz 提交于
Don't do blocking disk I/O under mutex blocking scrape and eval
-
由 Alin Sinpalean 提交于
Take the fdatasync() syscall out from under mutex lock to (mostly) avoid scrape and eval from blocking on disk I/O.
-