title | date | summary | tags | ||
---|---|---|---|---|---|
Weekly update (January 09 ~ January 22, 2017) |
2017-01-24 |
Last two weeks, we landed 87 PRs in the TiDB repositories and 50 PRs in the TiKV repositories. |
|
Last two weeks, we landed 87 PRs in the TiDB repositories.
-
Add some metrics: #2417, #2419, #2435, #2451, #2460, [#2477])(pingcap/tidb#2477)
-
Set a limitation about the key-value count and size in a single transaction.
-
Support
Rename Table
andAlter Table Rename Table
statement.
-
Change the argument name for displaying version from
v
toV
: keep consistent with pd/tikv.
Last two weeks, We landed 50 PRs in the TiKV repositories.
-
Set a limitation for the max size of the raft entry.
-
Make disabling manual compaction configurable.
-
Add fail reason tag in raft request and coprocessor metrics.
-
Add snapshot KV count and size metrics.
-
Add count of written keys metric.
-
Add raft log lag metric.
-
Add thread CPU collector.
-
Support
compaction_readahead_size
for spinning disk. -
Support getting region with key in
pd-ctl
. -
Statistic region write key-value count and bytes.
-
Fix a pitfall when use different context package.
-
Count KV count in other CF when check splitting.
-
Avoid submitting secondary key if primary key is locked.
-
Use disk size as storage capacity when capacity is set to 0.
-
Check locked for
ResolveLock
in latch.
-
Check promotable when handle raft MsgTimeoutNow.
-
Unify logger.
-
Resume paused follower when receive
MsgHeartbeatResp
message. -
Reset the pending state when add node.
-
Allow running one GC command at the same time.
-
Guarantee replicas are safe if possible.
-
Keep leader alive before initialization.
-
Log slow request for scheduler.
-
Record scan efficiency for coprocessor.
-
Make raft log GC not depend on PeerStorage.
-
Accelerate ticks for new split region leader.