Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

c-deps: bump rocksdb for multiple backported PRs #37172

Merged
merged 1 commit into from
Apr 27, 2019

Conversation

ajkr
Copy link
Contributor

@ajkr ajkr commented Apr 27, 2019

Includes the following changes, all of which have landed upstream.

Also made the RocksDB changes that we decided in #34897:

Release note: None

Includes the following changes, all of which have landed upstream.

- cockroachdb/rocksdb#27: "ldb: set `total_order_seek` for scans"
- cockroachdb/rocksdb#28: "Fix cockroachdb#3840: only `SyncClosedLogs` for multiple CFs"
- cockroachdb/rocksdb#29: "Optionally wait on bytes_per_sync to smooth I/O"
- cockroachdb/rocksdb#30: "Option string/map/file can set env from object registry"

Also made the RocksDB changes that we decided in cockroachdb#34897:

- Do not sync WAL before installing flush result. This is achieved by backporting cockroachdb/rocksdb#28; no configuration change is necessary.
- Do not sync WAL ever for temp stores. This is achieved by setting `wal_bytes_per_sync = 0`.
- Limit size of final syncs when generating SSTs. This is achieved by backporting cockroachdb/rocksdb#29 and turning it on with `strict_bytes_per_sync = true`.

Release note: None
@ajkr ajkr requested review from a team April 27, 2019 00:10
@cockroach-teamcity
Copy link
Member

This change is Reviewable

Copy link
Collaborator

@petermattis petermattis left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

options.strict_bytes_per_sync = true;
// Do not sync the WAL periodically. We sync it every write already by calling
// `FlushWAL(true)` on non-temp stores. On the temp store we do not intend to
// sync WAL ever, so setting it to zero is fine there too.
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While we don't intend to sync the WAL on temp-stores, I wonder if not syncing can run into problems where the WAL dirty data forces the OS to flush. Hopefully we're deleting the WAL before that happens, though. Ok, I just talked myself into being ok with this.

@ajkr
Copy link
Contributor Author

ajkr commented Apr 27, 2019

bors r+

craig bot pushed a commit that referenced this pull request Apr 27, 2019
37172: c-deps: bump rocksdb for multiple backported PRs r=ajkr a=ajkr

Includes the following changes, all of which have landed upstream.

- cockroachdb/rocksdb#27: "ldb: set `total_order_seek` for scans"
- cockroachdb/rocksdb#28: "Fix #3840: only `SyncClosedLogs` for multiple CFs"
- cockroachdb/rocksdb#29: "Optionally wait on bytes_per_sync to smooth I/O"
- cockroachdb/rocksdb#30: "Option string/map/file can set env from object registry"

Also made the RocksDB changes that we decided in #34897:

- Do not sync WAL before installing flush result. This is achieved by backporting cockroachdb/rocksdb#28; no configuration change is necessary.
- Do not sync WAL ever for temp stores. This is achieved by setting `wal_bytes_per_sync = 0`.
- Limit size of final syncs when generating SSTs. This is achieved by backporting cockroachdb/rocksdb#29 and turning it on with `strict_bytes_per_sync = true`.

Release note: None

Co-authored-by: Andrew Kryczka <andrew.kryczka2@gmail.com>
@craig
Copy link
Contributor

craig bot commented Apr 27, 2019

Build succeeded

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Failed tests (11329):
3 participants