- 19 5月, 2021 4 次提交
-
-
由 quicksilver 提交于
Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com> Resolves: #5293
-
由 congqixia 提交于
Restore segment flush stream & statistic stream to last success pos Signed-off-by: NCongqi Xia <congqi.xia@zilliz.com>
-
由 XuanYang-cn 提交于
Signed-off-by: Nyangxuan <xuan.yang@zilliz.com>
-
由 ThreadDao 提交于
- add cases for creating collections - change unique str from var to function - improve coding standards See also: #5273 #5285 Signed-off-by: ThreadDao yufen.zong@zilliz.com
-
- 18 5月, 2021 9 次提交
-
-
由 XuanYang-cn 提交于
Before this PR, DataNode considered auto-flush a valid flush complete. It's wrong. So I open this PR to correct this behaviour in DataNode. Now binlog paths from auto-flush will be buffered in replica, waiting until the manul flush to save into etcd all together. See also: #5220, #5268 A follow up job of #5271 Signed-off-by: Nyangxuan <xuan.yang@zilliz.com>
-
由 sunby 提交于
This document includes changes about dataservice and new features to support the recovery after it crashed. Signed-off-by: Nsunby <bingyi.sun@zilliz.com>
-
由 godchen 提交于
Add exclusive parameter. Issue #5174 Signed-off-by: Ngodchen <qingxiang.chen@zilliz.com>
-
由 neza2017 提交于
query base on timeline, add timestamp on these queries: - ShowCollections - HasCollection - DescribeCollection Resolves: #5219 Signed-off-by: Nyefu.chen <yefu.chen@zilliz.com>
-
由 XuanYang-cn 提交于
Auto-flush of a segment is not considered flush-completed. So we need to buffer binlog paths generated by auto-flush. See also: #5268 #5220 Signed-off-by: Nyangxuan <xuan.yang@zilliz.com>
-
由 neza2017 提交于
let meta support snapshot so collection could query meta with timestamp Resolves: #5219 Signed-off-by: Nyefu.chen <yefu.chen@zilliz.com>
-
由 Xiangyu Wang 提交于
Signed-off-by: NXiangyu Wang <xiangyu.wang@zilliz.com>
-
由 godchen 提交于
Add session util. Issue #5174 Signed-off-by: Ngodchen <qingxiang.chen@zilliz.com>
-
由 XuanYang-cn 提交于
Signed-off-by: Nyangxuan <xuan.yang@zilliz.com> The design procedure can be seen: #5220
-
- 17 5月, 2021 5 次提交
-
-
由 Cai Yudong 提交于
Updata PhysicalChannelNames and VirtualChannelNames based on ShardsNum when create collection, and update collVChan2Chan map. Resolves: #5240 Signed-off-by: Nyudong.cai <yudong.cai@zilliz.com>
-
由 ThreadDao 提交于
- update get_connection to create_connection - add case level for collection - optimize collection cases Signed-off-by: ThreadDao yufen.zong@zilliz.com
-
由 quicksilver 提交于
* Change volumes with pytest docker-compose Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com> * Fix run pytest bug Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 quicksilver 提交于
Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 ThreadDao 提交于
Add a part of the test case for the Collection interface See also: #5224 #5231 Signed-off-by: ThreadDao yufen.zong@zilliz.com
-
- 15 5月, 2021 4 次提交
-
-
由 Cai Yudong 提交于
Remove CreateIndexTask struct and indexTaskQueue Resolves: #5228 Signed-off-by: Nyudong.cai <yudong.cai@zilliz.com>
-
由 Xiangyu Wang 提交于
Signed-off-by: NXiangyu Wang <xiangyu.wang@zilliz.com>
-
由 quicksilver 提交于
Signed-off-by: jeffoverflow zhifeng.zhang@zilliz.com
-
由 neza2017 提交于
just remove standalone and milvus from $GOPATH/bin, not every thing, on command of make clean Signed-off-by: Nyefu.chen <yefu.chen@zilliz.com>
-
- 14 5月, 2021 11 次提交
-
-
由 quicksilver 提交于
Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 Cai Yudong 提交于
Resend unsuccessful ddMsg when master start Signed-off-by: Nyudong.cai <yudong.cai@zilliz.com>
-
由 quicksilver 提交于
* Deny jenkins slave pod to remain active for a while Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 quicksilver 提交于
Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 godchen 提交于
Add session util doc. Signed-off-by: Ngodchen <qingxiang.chen@zilliz.com>
-
由 XuanYang-cn 提交于
- FlushMsg proto message is no longer in use - We don't depend on "github.com/prometheus/common/log" Signed-off-by: yangxuan xuan.yang@zilliz.com
-
由 XuanYang-cn 提交于
* add datanode design doc Signed-off-by: Nyangxuan <xuan.yang@zilliz.com>
-
由 neza2017 提交于
design doc about master service recover on power failure Signed-off-by: Nyefu.chen <yefu.chen@zilliz.com>
-
由 yukun 提交于
GetEntityByID needs the flushed segment to be sorted by RowID field, then do binary search to get the target id and entities. See also: #5177 Signed-off-by: fishpenguin kun.yu@zilliz.com
-
由 godchen 提交于
Add service registration. Part of Issue #5174. Signed-off-by: Ngodchen <qingxiang.chen@zilliz.com>
-
由 紫晴 提交于
Signed-off-by: Nwangting0128 <ting.wang@zilliz.com>
-
- 13 5月, 2021 2 次提交
-
-
由 紫晴 提交于
update test code
-
由 quicksilver 提交于
Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com> Delete milvus PVC does not affect E2E Test workflow
-
- 12 5月, 2021 3 次提交
-
-
由 Cai Yudong 提交于
Save ddmsg info and type into ETCD to support re-send ddmsg when system restart or recover from failure (#5172) Signed-off-by: Nyudong.cai <yudong.cai@zilliz.com>
-
由 Xiangyu Wang 提交于
Signed-off-by: Nscsven <wxyucs@gmail.com>
-
由 yanliang567 提交于
Signed-off-by: Nyanliang567 <yanliang.qiao@zilliz.com>
-
- 11 5月, 2021 2 次提交
-
-
由 quicksilver 提交于
* Add publish-krte-images.yaml Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com> * Update publish-krte-images.yaml Signed-off-by: Nquicksilver <zhifeng.zhang@zilliz.com>
-
由 紫晴 提交于
Signed-off-by: Nwangting0128 <ting.wang@zilliz.com>
-