Matt Keller
19a28c8efd
storage: Fixed backend test
...
./backend_test.go:23: multiple-value batchTx.UnsafeRange() in single-value context
2015-08-27 15:20:29 -04:00
Matt Keller
32372e1d70
raft: Fixed a test misassumption
...
network_test.go:56: total = 59.22354ms, want > 50ms
59 is > 50, but the equation added 10 to the right side
2015-08-27 15:15:34 -04:00
Xiang Li
25c87f13fd
Merge pull request #3354 from mx2323/faq
...
add faq documentation
2015-08-26 16:36:04 -07:00
Michael Xu
8f3ea5ebed
doc: add faq documentation
2015-08-26 16:34:52 -07:00
Yicheng Qin
59a5a7e309
Merge pull request #3368 from yichengq/storage-test
...
add unit tests for storage
2015-08-26 15:32:02 -07:00
Yicheng Qin
0d38c13990
storage: use temp path to handle test file
2015-08-26 15:01:41 -07:00
Yicheng Qin
2d01eb4e11
storage: add tests for kvstore_compaction
2015-08-26 15:01:13 -07:00
Yicheng Qin
f38778160d
Merge pull request #3376 from yichengq/connection-down
...
etcdserver: specify request timeout error due to connection down
2015-08-26 13:09:30 -07:00
Yicheng Qin
0813139140
storage: add more tests for index
2015-08-26 12:53:30 -07:00
Yicheng Qin
3723f01b48
storage: add more unit tests for keyIndex
2015-08-26 12:53:30 -07:00
Yicheng Qin
ad8a291dc1
storage: return error when tombstone on new generation
...
It is not allowed to put tombstone on an empty generation.
2015-08-26 12:53:30 -07:00
Yicheng Qin
ffa87f9678
storage: fix the comment in generation.walk
2015-08-26 12:53:30 -07:00
Yicheng Qin
8f6bf029f8
etcdserver: specify request timeout error due to connection lost
...
It specifies request timeout error possibly caused by connection lost,
and print out better log for user to understand.
It handles two cases:
1. the leader cannot connect to majority of cluster.
2. the connection between follower and leader is down for a while,
and it losts proposals.
log format:
```
20:04:19 etcd3 | 2015-08-25 20:04:19.368126 E | etcdhttp: etcdserver:
request timed out, possibly due to connection lost
20:04:19 etcd3 | 2015-08-25 20:04:19.368227 E | etcdhttp: etcdserver:
request timed out, possibly due to connection lost
```
2015-08-26 12:38:37 -07:00
Yicheng Qin
76db9747f8
Merge pull request #3377 from yichengq/tls-info-string
...
pkg/transport: print ClientCertAuth in TLSInfo.String()
2015-08-25 22:45:10 -07:00
Yicheng Qin
45bb88069b
Merge pull request #3378 from yichengq/set-late
...
etcdmain: check error before assigning peer transport
2015-08-25 22:38:36 -07:00
Yicheng Qin
58455a2ae4
etcdmain: check error before assigning peer transport
...
Or it may panic when new transport fails, e.g., TLS info is invalid.
2015-08-25 22:04:26 -07:00
Yicheng Qin
57e88465bf
pkg/transport: print ClientCertAuth in TLSInfo.String()
...
It is good to print it in debug output:
```
21:56:12 etcd1 | 2015-08-25 21:56:12.162406 I | etcdmain: peerTLS: cert
= certs/etcd1.pem, key = certs/etcd1-key.pem, ca = , trusted-ca =
certs/ca.pem, client-cert-auth = true
```
2015-08-25 21:53:52 -07:00
Yicheng Qin
6250fed8a8
Merge pull request #3096 from philips/tls-info-debug
...
pkg/transport: include debug output for trusted-ca
2015-08-25 20:08:19 -07:00
Xiang Li
008f988f6b
Merge pull request #3375 from xiang90/doc
...
doc: add evn variable name to configuration.md
2015-08-25 14:48:35 -07:00
Yicheng Qin
2b58da1699
Merge pull request #3374 from yichengq/gomaxprocs
...
etcdmain: change default GOMAXPROCS when compiling in go1.5
2015-08-25 14:48:00 -07:00
Xiang Li
35a0459cc8
doc: add evn variable name to configuration.md
2015-08-25 14:35:15 -07:00
Xiang Li
32ab3f6931
Merge pull request #3372 from xiang90/doc
...
improve clustering.md doc
2015-08-25 14:04:30 -07:00
Xiang Li
c30c85898e
doc: add explanation for client urls
2015-08-25 13:46:27 -07:00
Yicheng Qin
2ac9a329ab
etcdmain: stop setting GOMAXPROCS explicitly
...
We always want to use GOMAXPROCS() as the way go parses it. When in go1.4, we
want to expose GOMAXPROCS value, so we set GOMAXPROCS explicitly as the
way go 1.4 does and print it out.
But it becomes a problem when go 1.5 changes the way to set GOMAXPROCS.
Fix the problem by stop setting GOMAXPROCS and get its value directly.
Due to this change, it sets default GOMAXPROCS to the
number of CPUs available when compiling in go 1.5, which matches how go 1.5 works:
https://docs.google.com/document/d/1At2Ls5_fhJQ59kDK2DFVhFu3g5mATSXqqV5QrxinasI/edit
This is a behavior change in etcd 2.2.
2015-08-25 13:38:16 -07:00
Yicheng Qin
a4285ef5c9
Merge pull request #3367 from MSamman/master
...
etcdserver: handle malformed basic auth
2015-08-25 13:12:48 -07:00
Mohammad Samman
e2e002f94e
etcdserver: handle malformed basic auth
...
return insufficient credentials if basic auth header is malformed
Fixes #3280
2015-08-25 12:37:24 -07:00
Xiang Li
7bd558b2e0
Merge pull request #3373 from ecnahc515/add_report_bugs_contributing
...
Contributing: Link to reporting bugs doc
2015-08-25 12:17:06 -07:00
Chance Zibolski
ad843341a9
Contributing: Link to reporting bugs doc
2015-08-25 12:15:03 -07:00
Xiang Li
f56c5455f3
doc: mention reconfiguration design in clustering.md
2015-08-25 11:22:08 -07:00
Xiang Li
986f354694
Merge pull request #3371 from xiang90/bolt
...
Godeps: update bolt dependency
2015-08-25 11:17:14 -07:00
Xiang Li
e8f40b0412
storage/backend: add commitAndStop
...
After the upgrade of boltdb, db.Close waits for all txn to finish.
CommitAndStop commits the current txn and stop creating new ones.
2015-08-25 10:57:25 -07:00
Xiang Li
8738a88fae
Godeps: update bolt dependency
2015-08-25 10:39:29 -07:00
Yicheng Qin
2d06f6b371
Merge pull request #3362 from yichengq/rafthttp-cancel
...
rafthttp: always cancel in-flight request when stop streamReader
2015-08-25 09:26:46 -07:00
Yicheng Qin
61a75b3d48
rafthttp: always cancel in-flight request when pipeline.send
...
This fits the way for go1.5 to cancel request.
2015-08-25 09:07:49 -07:00
Yicheng Qin
27b9963959
client: always cancel in-flight request when do request
...
This fits the way for go1.5 to cancel request.
2015-08-25 09:04:58 -07:00
Yicheng Qin
ece39c9462
proxy: always cancel in-flight request
...
This fits the way for go1.5 to cancel request.
2015-08-25 08:59:59 -07:00
Yicheng Qin
6fc638673c
rafthttp: return err if stopped before setting cancel in dial()
...
The original workflow may fail to cancel if stop() cancels the finished
request just before dial() assigning a new cancel. This commit checks
streamReader status before setting cancel to avoid this problem.
It is tested at travis for 300 times. go 1.5 always works well, while
go 1.4 fails to stop once.
2015-08-25 08:59:12 -07:00
Yicheng Qin
fc95ec0cc6
rafthttp: always cancel in-flight request when stop streamReader
...
This problem is totally fixed at 1.5.
go1.5 adds a Request.Cancel channel, which allows for "race free"
cancellation
(8b4278ffb7
).
Our implementation relies on it to always cancel in-flight request.
2015-08-25 08:54:13 -07:00
Yicheng Qin
0132b091d2
Merge pull request #3360 from yichengq/bench-3
...
*: add initial read benchmark for etcd v3
2015-08-25 07:58:30 -07:00
Yicheng Qin
3632a1b9b1
*: add initial read benchmark for etcd v3
...
It includes the initial read benchmark for etcd v3.
This is the first step to give some rough thoughts. I haven't digged
deeper to answer some questions, including why its performance is not
better than HTTP + json, why one put will cause performance downgrade.
2015-08-25 07:50:18 -07:00
Xiang Li
e3ef1d363a
Merge pull request #3366 from xiang90/v3_proto
...
update v3 proto and doc
2015-08-24 11:22:29 -07:00
Xiang Li
0cb45aee64
rfc: update v3 proto
2015-08-24 11:00:51 -07:00
Xiang Li
1cccbb5ebd
etcdserverpb: add comments for compaction
2015-08-24 10:52:54 -07:00
Xiang Li
3a60d490d1
storagepb: fix comment location
2015-08-24 10:42:16 -07:00
Xiang Li
4a5b94478e
etcdserverpb: update comment for txn request
2015-08-24 10:40:05 -07:00
Xiang Li
98ceb3cdbd
etcdserverpb: add more field into rangeResponse
2015-08-24 10:33:20 -07:00
Yicheng Qin
c7f10ed975
Merge pull request #3361 from yichengq/no-log
...
integration: only print critical log
2015-08-24 09:44:13 -07:00
Yicheng Qin
3702be476b
integration: only print critical log
...
This limits the logs printed out in integration test, so it will not
have log flood and help us read fatal log in travis.
2015-08-23 21:22:21 -07:00
Yicheng Qin
514c4371a9
Merge pull request #3359 from yichengq/storage-test
...
functional tests for storage package and some related fixes
2015-08-23 21:12:36 -07:00
Yicheng Qin
1e2b0acf6d
test: activate test for storage package
2015-08-23 20:59:06 -07:00