| a23109a | Gyu-Ho Lee | 13 January 2017, 19:29:12 UTC | version: bump to v3.0.16 | 13 January 2017, 19:29:12 UTC |
| 219a4e9 | Anthony Romano | 16 December 2016, 08:28:17 UTC | clientv3: don't reset keepalive stream on grant failure Was triggering cancelation errors on outstanding KeepAlives if Grant had to retry. | 13 January 2017, 19:28:17 UTC |
| 3d05063 | Anthony Romano | 16 December 2016, 08:21:07 UTC | v3api, rpctypes: add ErrTimeoutDueToConnectionLost Lack of GRPC code was causing this to look like a halting error to the client. | 13 January 2017, 19:27:56 UTC |
| 9c66ed2 | Anthony Romano | 12 January 2017, 23:53:47 UTC | clientv3: don't reset stream on keepaliveonce or revoke failure Would cause the keepalive loop to cancel out. Fixes #7082 | 13 January 2017, 18:42:01 UTC |
| a9e2d3d | Gyu-Ho Lee | 07 December 2016, 18:17:28 UTC | *: remove 'tools/etcd-top' to drop pcap.h | 07 December 2016, 18:34:34 UTC |
| 41e329c | Gyu-Ho Lee | 07 December 2016, 18:15:34 UTC | *: drop breaking-changes from master branch | 07 December 2016, 18:34:28 UTC |
| 3a8b524 | Gyu-Ho Lee | 07 December 2016, 18:07:43 UTC | travis, test: use Go 1.6.4, skip 'gosimple' | 07 December 2016, 18:28:53 UTC |
| 11668f5 | Anthony Romano | 03 December 2016, 06:14:44 UTC | integration: use RequireLeader for TestV3LeaseFailover Giving Renew() the default request timeout causes TestV3LeaseFailover to miss its timing constraints. Since it only needs to wait until the leader recognizes the leader is lost, use RequireLeader to cancel the keepalive stream before the request times out. | 07 December 2016, 18:06:28 UTC |
| 7ceca7e | Anthony Romano | 02 December 2016, 22:20:00 UTC | clientv3/integration: test lease keepalive works following quorum loss | 07 December 2016, 18:06:28 UTC |
| 395bd23 | Anthony Romano | 02 December 2016, 21:19:20 UTC | v3rpc, etcdserver, leasehttp: ctxize Renew with request timeout Would retry a few times before returning a not primary error that the client should never see. Instead, use proper timeouts and then return a request timeout error on failure. Fixes #6922 | 07 December 2016, 18:06:19 UTC |
| b357569 | Gyu-Ho Lee | 11 November 2016, 19:17:31 UTC | version: bump to v3.0.15+git | 11 November 2016, 19:17:31 UTC |
| fc00305 | Gyu-Ho Lee | 10 November 2016, 21:12:43 UTC | version: bump to v3.0.15 | 10 November 2016, 21:12:43 UTC |
| f322fe7 | Gyu-Ho Lee | 08 November 2016, 19:46:52 UTC | clientv3, ctlv3: document range end requirement | 10 November 2016, 21:10:18 UTC |
| 049fcd3 | Gyu-Ho Lee | 08 November 2016, 18:57:01 UTC | integration: test wrong watcher range | 10 November 2016, 21:09:13 UTC |
| 1b702e7 | Gyu-Ho Lee | 08 November 2016, 18:49:44 UTC | mvcc: return -1 for wrong watcher range key >= end Fix https://github.com/coreos/etcd/issues/6819. | 10 November 2016, 21:08:51 UTC |
| b87190d | Anthony Romano | 07 November 2016, 20:04:41 UTC | integration: test canceling a watcher on disconnected stream | 10 November 2016, 21:07:24 UTC |
| 83b493f | Anthony Romano | 07 November 2016, 21:03:47 UTC | clientv3: let watchers cancel when reconnecting | 10 November 2016, 21:06:47 UTC |
| 9b69cbd | Gyu-Ho Lee | 04 November 2016, 20:06:36 UTC | version: bump to v3.0.14+git | 04 November 2016, 20:06:36 UTC |
| 8a37349 | Gyu-Ho Lee | 04 November 2016, 17:54:14 UTC | version: bump to v3.0.14 | 04 November 2016, 17:54:14 UTC |
| 9a0e4df | Xiang Li | 03 November 2016, 02:36:07 UTC | ctlv3: fix migration | 03 November 2016, 16:47:41 UTC |
| f60469a | Timothy St. Clair | 02 November 2016, 23:10:30 UTC | ctlv3: Add a no-ttl flag to etcdctl migrate to discard keys on transform. | 03 November 2016, 16:47:39 UTC |
| 932370d | Gyu-Ho Lee | 24 October 2016, 18:22:50 UTC | version: bump to v3.0.13+git | 24 October 2016, 18:22:50 UTC |
| c99d0d4 | Gyu-Ho Lee | 24 October 2016, 18:04:43 UTC | version: bump to v3.0.13 | 24 October 2016, 18:04:43 UTC |
| d78216f | Gyu-Ho Lee | 24 October 2016, 18:04:13 UTC | e2e: remove 'ctlV3GetFailPerm' | 24 October 2016, 18:04:13 UTC |
| c05c027 | Hongchao Deng | 20 October 2016, 17:03:44 UTC | etcdctl: fix migrate in outputing client.Node to json Using printf will try to parse the string and replace special characters. In migrate code, we want to just output the raw json string of client.Node. For example, Printf("%\\") => %!\(MISSING) Print("%\\") => %\ Thus, we should use print instead. | 20 October 2016, 17:51:16 UTC |
| 3fd64f9 | Gyu-Ho Lee | 12 October 2016, 20:59:27 UTC | auth: fix return type on 'hasRootRole' | 12 October 2016, 20:59:27 UTC |
| f935290 | Xiang Li | 12 October 2016, 17:42:55 UTC | mvcc: fix rev inconsistency Try: ./etcdctl put foo bar ./etcdctl del foo ./etcdctl compact 3 restart etcd ./etcdctl get foo mvcc: required revision has been compacted The error is unexpected when range over the head revision. Internally, we incorrectly set current revision smaller than the compacted revision when we remove all keys around compacted revision. This commit fixes the issue by recovering the current revision at least to compacted revision. | 12 October 2016, 20:08:26 UTC |
| ca91f89 | Hitoshi Mitake | 06 September 2016, 02:32:25 UTC | auth, e2e, clientv3: the root role should be granted access to every key This commit changes the semantics of the root role. The role should be able to access to every key. Partially fixes https://github.com/coreos/etcd/issues/6355 | 11 October 2016, 19:19:46 UTC |
| fcbada7 | Gyu-Ho Lee | 11 October 2016, 19:15:58 UTC | Merge pull request #6622 from luxas/backport_arm_fixes Backport arm fixes | 11 October 2016, 19:15:58 UTC |
| fad9bdc | Jared Hulbert | 08 July 2016, 18:20:47 UTC | etcdserver: atomic access alignment Most fields accessed with sync/atomic functions are 64bit aligned, but a couple are not. This makes comments out of date and therefore misleading. Affected fields reordered, comments scrubbed and updated. | 11 October 2016, 08:48:43 UTC |
| 198ccb8 | Jared Hulbert | 08 July 2016, 18:13:53 UTC | raftpb: atomic access alignment The Entry struct has misaligned fields that are accessed atomically. The misalignment is caused by the EntryType enum which the Protocol Buffers spec forces to be a 32bit int. Moving the order of the fields without renumbering them in the .proto file seems to align the go structure without changing the wire format. | 11 October 2016, 08:48:43 UTC |
| dc5d5c6 | Jared Hulbert | 08 July 2016, 18:05:41 UTC | raft: atomic access alignment The relevant structures are properly aligned, however, there is no comment highlighting the need to keep it aligned as is present elsewhere in the codebase. Adding note to keep alignment, in line with similar comments in the codebase. | 11 October 2016, 08:48:43 UTC |
| f771eac | Gyu-Ho Lee | 07 October 2016, 23:42:12 UTC | version: bump to v3.0.12+git | 07 October 2016, 23:42:12 UTC |
| 2d1e2e8 | Gyu-Ho Lee | 07 October 2016, 22:14:25 UTC | version: bump to v3.0.12 | 07 October 2016, 22:14:25 UTC |
| 6412758 | Gyu-Ho Lee | 07 October 2016, 22:13:56 UTC | v3rpc: remove redundant locks | 07 October 2016, 22:13:56 UTC |
| 836c815 | Xiang Li | 07 July 2016, 22:01:05 UTC | v3rpc: lock progress and prevKV map correctly | 07 October 2016, 22:13:12 UTC |
| e406e6e | Gyu-Ho Lee | 07 October 2016, 21:23:09 UTC | etcdctl/ctlv3: add 'prev-kv' flag to watch command | 07 October 2016, 21:23:09 UTC |
| 2fa2c62 | Gyu-Ho Lee | 07 October 2016, 21:22:58 UTC | clientv3: add 'prevKV' field to watch request | 07 October 2016, 21:22:58 UTC |
| 2862c4f | Gyu-Ho Lee | 07 October 2016, 21:22:19 UTC | v3rpc: implement 'prev-kv' watch | 07 October 2016, 21:22:19 UTC |
| 6f89fbf | Gyu-Ho Lee | 07 October 2016, 21:21:39 UTC | etcdserver: use mvcc.WatchableKV for prev-kv watch | 07 October 2016, 21:22:00 UTC |
| 6ae7ec9 | Gyu-Ho Lee | 07 October 2016, 21:21:19 UTC | *: regenerate proto | 07 October 2016, 21:21:19 UTC |
| 4a35b1b | Gyu-Ho Lee | 07 October 2016, 21:20:46 UTC | etcdserverpb: add 'prev_kb' to WatchCreateRequest | 07 October 2016, 21:20:46 UTC |
| c859c97 | Gyu-Ho Lee | 07 October 2016, 21:19:59 UTC | mvccpb: add 'prev_kv' field | 07 October 2016, 21:19:59 UTC |
| a091c62 | Gyu-Ho Lee | 07 October 2016, 20:25:21 UTC | version: bump to v3.0.11+git | 07 October 2016, 20:25:21 UTC |
| 96de94a | Gyu-Ho Lee | 07 October 2016, 18:27:48 UTC | version: bump to v3.0.11 | 07 October 2016, 18:27:48 UTC |
| e9cd841 | Gyu-Ho Lee | 07 October 2016, 18:03:19 UTC | integration: add 'prevKV' to TestV3DeleteRange | 07 October 2016, 18:03:19 UTC |
| e37ede1 | Gyu-Ho Lee | 07 October 2016, 18:00:48 UTC | etcdserver: handle 'PrevKV' | 07 October 2016, 18:00:48 UTC |
| 4420a29 | Gyu-Ho Lee | 07 October 2016, 17:56:06 UTC | etcdctl/ctlv3: add 'prev-kv' flag | 07 October 2016, 17:56:06 UTC |
| 0544d4b | Gyu-Ho Lee | 07 October 2016, 17:54:45 UTC | clientv3: add WithPrevKV OpOption | 07 October 2016, 17:54:45 UTC |
| fe7379f | Gyu-Ho Lee | 07 October 2016, 17:51:01 UTC | clientv3: add Op.prevKV | 07 October 2016, 17:51:01 UTC |
| c76df50 | Gyu-Ho Lee | 07 October 2016, 17:47:47 UTC | *: update proto to add 'prev_kv' | 07 October 2016, 17:47:47 UTC |
| 3299cad | Xiang Li | 06 July 2016, 00:08:09 UTC | *: add put prevkv | 07 October 2016, 17:39:08 UTC |
| d9ab018 | Anthony Romano | 04 October 2016, 21:47:40 UTC | integration: test a canceled watch won't return a closing error | 05 October 2016, 21:19:36 UTC |
| e853451 | Anthony Romano | 04 October 2016, 23:09:50 UTC | clientv3: only return closing error to watcher if context is not canceled Fixes #6503 | 05 October 2016, 21:19:32 UTC |
| 1becf9d | Anthony Romano | 05 October 2016, 16:36:06 UTC | clientv3: fix race on watch initial revision The initial revision was being updated in the substream goroutine defer; this was racing with the resume path fetching the initial revision when the substream closes during resume. Instead, update the initial revision whenever the substream processes a new watch response. Since the substream cannot receive a watch response while it is resuming, the write to the initial revision is ordered to always happen after the resume read. Fixes #6586 | 05 October 2016, 17:56:36 UTC |
| 1a712cf | Anthony Romano | 28 September 2016, 23:47:33 UTC | clientv3: make IsProgressNotify() false on compact event and closed channel Fixes #6549 | 04 October 2016, 22:13:02 UTC |
| 023f335 | Gyu-Ho Lee | 28 September 2016, 18:03:24 UTC | wal: set PageWriter offset in file encoder | 04 October 2016, 22:12:47 UTC |
| bf0da78 | Gyu-Ho Lee | 28 September 2016, 16:41:24 UTC | pkg/ioutil: configure pageOffset in NewPageWriter | 04 October 2016, 22:12:46 UTC |
| e847385 | Anthony Romano | 23 September 2016, 23:40:29 UTC | integration: test canceling watchers when disconnected | 04 October 2016, 22:12:37 UTC |
| b836d18 | Anthony Romano | 23 September 2016, 23:47:25 UTC | clientv3: simplify watch synchronization Was more complicated than it needed to be and didn't really work in the first place. Restructured watcher registation to use a queue. | 04 October 2016, 22:12:18 UTC |
| 9b09229 | Gyu-Ho Lee | 23 September 2016, 18:13:45 UTC | version: bump to v3.0.10+git | 23 September 2016, 18:13:45 UTC |
| 546c0f7 | Gyu-Ho Lee | 23 September 2016, 17:49:03 UTC | version: bump to v3.0.10 | 23 September 2016, 17:49:03 UTC |
| adbad1c | sharat | 16 September 2016, 15:57:22 UTC | ctlv3: close snapshot file before rename (Windows) | 23 September 2016, 16:11:02 UTC |
| 273b986 | Anthony Romano | 20 September 2016, 23:51:58 UTC | clientv3: process closed watcherStreams in watcherGrpcStream run loop Was racing with Watch() when closing the grpc stream on no watchers. Fixes #6476 | 21 September 2016, 22:52:20 UTC |
| 5b20572 | Gyu-Ho Lee | 16 September 2016, 12:54:55 UTC | rafthttp: add v3.0.0 to supported streams | 16 September 2016, 12:54:55 UTC |
| fe900b0 | Anthony Romano | 15 September 2016, 22:10:23 UTC | version: bump to v3.0.9+git | 15 September 2016, 22:10:23 UTC |
| 494c012 | Anthony Romano | 15 September 2016, 19:56:33 UTC | version: bump to v3.0.9 | 15 September 2016, 19:56:33 UTC |
| 4abc381 | Anthony Romano | 07 September 2016, 00:13:05 UTC | clientv3: drain buffered WatchResponses before resuming Otherwise, the watcherStream can receive WatchResponses in the middle of a resume, corrupting the stream. Fixes #6364 | 15 September 2016, 19:38:15 UTC |
| 73c8fda | Anthony Romano | 15 September 2016, 18:45:37 UTC | integration: fix compilation for backported Election test | 15 September 2016, 18:45:37 UTC |
| ee27174 | sharat | 08 September 2016, 20:28:33 UTC | ctlv3: fix line parsing for Windows | 15 September 2016, 18:25:53 UTC |
| 2435eb9 | Xiang Li | 14 September 2016, 05:52:16 UTC | clientv3: balancer panics when call up after close Fix the issue by adding a simple guard varable. | 15 September 2016, 09:46:26 UTC |
| 8fb533d | Anthony Romano | 13 September 2016, 02:49:22 UTC | embed: warn on domain name in listener | 15 September 2016, 09:46:19 UTC |
| 2f0f5ac | Anthony Romano | 13 September 2016, 02:45:35 UTC | Revert "Merge pull request #6365 from heyitsanthony/fix-dns-bind" This reverts commit af5ab7b3515cf023af7af94bc6e9063945802be0, reversing changes made to da6a0f0594f352539349013336614e08033a7eb3. | 15 September 2016, 09:43:46 UTC |
| 9ab811d | Jason E. Aten | 12 September 2016, 02:49:06 UTC | auth: fix range handling bugs. Test 15, counting from zero, in TestGetMergedPerms in etcd/auth/range_perm_cache_test.go, was trying incorrectly assert that [a, b) merged with [b, "") should be [a, b). Added a test specifically for this. This patch fixes the incorrect larger test and the bugs in the code that it was hiding. Fixes #6359 | 15 September 2016, 09:41:56 UTC |
| e0a99fb | Anthony Romano | 09 September 2016, 22:54:56 UTC | version: bump to v3.0.8+git | 09 September 2016, 22:56:31 UTC |
| d40982f | Anthony Romano | 09 September 2016, 20:14:44 UTC | version: bump to v3.0.8 | 09 September 2016, 20:14:44 UTC |
| fe3a1cc | Gyu-Ho Lee | 09 September 2016, 00:11:25 UTC | wal: fix error type | 09 September 2016, 00:11:25 UTC |
| 7071370 | Gyu-Ho Lee | 09 September 2016, 00:07:48 UTC | wal: fix err shadowing (go vet) | 09 September 2016, 00:07:48 UTC |
| 0054e7e | Xiang Li | 07 September 2016, 03:12:35 UTC | etcdctl: restore should create a snapshot Restore should create a snasphot. So the new db file can be sent to newly joined member. | 09 September 2016, 00:03:51 UTC |
| 97f718b | Anthony Romano | 08 September 2016, 06:24:29 UTC | fileutil: windows OpenDir Windows needs to open a directory with write access to fsync but the go runtime won't open directories that way. | 09 September 2016, 00:01:56 UTC |
| 202da92 | Anthony Romano | 07 September 2016, 23:40:33 UTC | wal: fsync directory after wal file rename Fixes #6368 | 09 September 2016, 00:01:49 UTC |
| 6e83ec0 | Anthony Romano | 06 September 2016, 22:30:51 UTC | etcdmain: reject binding listeners to domain names Fixes #6336 | 06 September 2016, 23:08:35 UTC |
| 5c44cdf | Jason E. Aten | 03 September 2016, 17:28:32 UTC | etcdctl/ctlv3: don't crash when we should prompt for pw. when 'etcdctl --user name get blah' is invoked to prompt for password, don't panic. addresses the segfault part of #6343 | 04 September 2016, 00:02:50 UTC |
| 09a239f | Anthony Romano | 31 August 2016, 23:59:26 UTC | e2e: add quoted key/value to txn test | 04 September 2016, 00:02:47 UTC |
| 3faff8b | Anthony Romano | 31 August 2016, 20:54:45 UTC | etcdctl: fix quoted string handling in txn and watch Fixes #6315 | 04 September 2016, 00:02:28 UTC |
| 2345fda | Anthony Romano | 31 August 2016, 23:41:06 UTC | version: bump to v3.0.7+git | 31 August 2016, 23:41:06 UTC |
| 5695120 | Gyu-Ho Lee | 31 August 2016, 16:49:24 UTC | version: bump to v3.0.7 | 31 August 2016, 16:49:24 UTC |
| 183293e | Gyu-Ho Lee | 31 August 2016, 16:48:11 UTC | wal: lowercase segmentSizeBytes | 31 August 2016, 16:48:30 UTC |
| 4b48876 | Jason E. Aten | 28 August 2016, 05:58:57 UTC | clientv3/concurrency: allow election on prefixes of keys. After winning an election or obtaining a lock, we auto-append a slash after the provided key prefix. This avoids the previous deadlock due to waiting on the wrong key. Fixes #6278 Conflicts: clientv3/concurrency/election.go clientv3/concurrency/mutex.go | 31 August 2016, 16:46:05 UTC |
| 5089bf5 | Aaron Lehmann | 23 August 2016, 21:48:07 UTC | wal: hold file lock while renaming WAL directory on non-Windows Windows requires this lock to be released before the directory is renamed. But on unix-like operating systems, releasing the lock and trying to reacquire it immediately can be flaky if a process is forked around the same time. The file descriptors are marked as close-on-exec by the Go runtime, but there is a window between the fork and exec where another process will be holding the lock. | 31 August 2016, 16:39:57 UTC |
| 480a347 | Anthony Romano | 30 August 2016, 20:40:47 UTC | wal: use page buffered writer for writing records Forces torn writes to only happen on sector boundaries. Fixes #6271 | 31 August 2016, 04:06:36 UTC |
| 59e560c | Anthony Romano | 30 August 2016, 20:22:23 UTC | ioutil: add page buffered writer A buffered writer that only writes full pages or when explicitly flushed. | 31 August 2016, 04:06:33 UTC |
| 0bd9bea | Xiang Li | 30 August 2016, 17:43:36 UTC | etcdserver: allow zero kv index for cluster upgrade If a user upgrades etcd from 2.3.x to 3.0 and shutdown the cluster immediately without triggering any new backend writes, then the consistent index in backend would be zero. The user cannot restart etcdserver due to today's strick index match checking. We now have to lose this a bit for this case. | 31 August 2016, 04:05:20 UTC |
| bd7581a | Anthony Romano | 25 August 2016, 20:23:36 UTC | wal: zero out wal tail past its first zero record Whenever the WAL is opened for writes, it should write zeroes to its tail starting from the first zero record. Otherwise, if there are entries past the first zero record due to a torn write, any new writes that overlap the old entries will lead to a garbage record on the tail and cause a CRC mismatch. | 26 August 2016, 21:27:53 UTC |
| db378c3 | Anthony Romano | 25 August 2016, 20:16:00 UTC | wal: test for truncation on torn writes | 26 August 2016, 21:27:51 UTC |
| 2374016 | Anthony Romano | 25 August 2016, 20:15:13 UTC | fileutil: add ZeroToEnd for zeroing files | 26 August 2016, 21:27:49 UTC |
| 96422a9 | Anthony Romano | 24 August 2016, 01:03:03 UTC | discovery: reject IP address records in SRVGetCluster Was incorrectly trimming the trailing '.' from the target; this in turn caused the etcd server to accept any SRV record with an IP target instead of only targets with A records. | 24 August 2016, 16:14:47 UTC |
| 6fd996f | Gyu-Ho Lee | 19 August 2016, 19:38:13 UTC | version: bump to v3.0.6+git | 19 August 2016, 19:38:13 UTC |
| 9efa00d | Gyu-Ho Lee | 19 August 2016, 19:03:02 UTC | version: bump to v3.0.6 | 19 August 2016, 19:03:02 UTC |
| 72d30f4 | Xiang Li | 05 August 2016, 03:39:32 UTC | *: minor cleanup for lease | 19 August 2016, 18:53:38 UTC |
