Commit Graph

11936 Commits (195744aea64bfa2d5f9e28e536522e9a58c036bb)

Author SHA1 Message Date
Anthony Romano 87d16af2e2 embed: use transport TLS listener for client listener for CRLs 2017-06-19 15:23:41 -07:00
Anthony Romano 7d7d1ae6a0 etcdmain: configure CRL file through command line 2017-06-19 15:23:41 -07:00
Anthony Romano 322976bedc transport: CRL checking 2017-06-19 15:23:41 -07:00
Xiang Li a65e3c69a6 Merge pull request #8122 from yudai/fast_fail_proxy
grpcproxy: Disable fast fail on lease grant call to cluster
2017-06-19 15:04:25 -07:00
Anthony Romano 66f553a96b Merge pull request #8127 from heyitsanthony/fix-restore
mvcc: restore into tree index with one key index
2017-06-19 12:58:18 -07:00
Gyu-Ho Lee 8f8f550443 test: 'FAIL' on release binary download failure
I see CI is failing to download release binaries
but exit code doesn't trigger CI job failure.

We need 'FAIL' string.

Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-19 12:52:02 -07:00
Anthony Romano 51a568aa81 mvcc: restore into tree index with one key index
Clobbering the mvcc kvindex with new keyIndexes for each restore
chunk would cause index corruption by dropping historical information.
2017-06-19 12:04:01 -07:00
Anthony Romano 02164874d9 mvcc: test restore and deletes with small chunk sizes 2017-06-19 12:04:01 -07:00
Anthony Romano 45fbac5544 Merge pull request #8025 from heyitsanthony/txn-cmp-range
api: txn comparisons on ranges
2017-06-18 11:11:43 -07:00
Iwasaki Yudai df2cc4bc8c grpcproxy: Disable fast fail on lease grant call to cluster
Problem Observed
----------------

When there is no etcd process behind the proxy,
clients repeat resending lease grant requests without delay.
This behavior can cause abnormal resource consumption on CPU/RAM and
network.

Problem Detail
--------------

`LeaseGrant()` uses a bare protobuf client to forward requests.
However, it doesn't use `grpc.FailFast(false)`, which means the method returns
an `Unavailable` error immediately when no etcd process is available.
In clientv3, `Unavailable` errors are not considered the "Halt" error,
and library retries the request without delay.
Both clients and the proxy consume much CPU cycles to process retry requests.

Resolution
----------

Add `grpc.FailFast(false))` to `LeaseGrant()` of the `leaseProxy`.
This makes the proxy not to return immediately when no etcd process is
available. Clients will simply timeout requests instead.
2017-06-16 15:09:05 -07:00
Anthony Romano e475a4ea71 Merge pull request #8120 from heyitsanthony/restore-set-size-metric
mvcc: set db size metric on restore
2017-06-16 12:37:08 -07:00
Anthony Romano 8f34d0c8b6 clientv3/integration: test compare on range 2017-06-16 12:13:27 -07:00
Anthony Romano 7ff6e62c56 namespace: prefix comparison range_end 2017-06-16 12:13:27 -07:00
Anthony Romano aeb2dc03aa grpcproxy: invalidate cache on comparison range 2017-06-16 12:13:27 -07:00
Anthony Romano fcf1abd23b clientv3: compare helper functions to set range/prefix 2017-06-16 12:13:27 -07:00
Anthony Romano fafb054624 integration: test txn range comparisons 2017-06-16 12:13:27 -07:00
Anthony Romano 8d7c29c732 etcdserver, etcdserverpb: Txn.Compare range_end support 2017-06-16 12:13:27 -07:00
Anthony Romano 7f149d8fb6 mvcc: set db size metric on restore
Fixes #8080
2017-06-16 11:27:34 -07:00
Anthony Romano a825709940 integration: test mvcc db size metric is set on restore 2017-06-16 11:27:07 -07:00
Anthony Romano 1acc8090e3 Merge pull request #8110 from heyitsanthony/fix-test-sync-timeout
etcdserver: use RecorderStream for TestSyncTimeout to avoid missing action
2017-06-15 20:49:10 -07:00
Anthony Romano e962b0c849 Merge pull request #7909 from heyitsanthony/unptr-cfg
etcdserver, embed, integration: don't use pointer for ServerConfig
2017-06-15 20:47:30 -07:00
Xiang Li 44a6c2121b Merge pull request #7999 from hexfusion/grpc-gateway-auth
auth: support "authorization" token for grpc-gateway
2017-06-15 19:22:00 -07:00
Anthony Romano 8fa96cb303 Merge pull request #8113 from heyitsanthony/code-of-conduct
*: add code of conduct
2017-06-15 19:18:24 -07:00
Anthony Romano 42584f84b4 *: add code of conduct
github community insights complains there isn't one
2017-06-15 17:04:45 -07:00
Anthony Romano 03ab4d9cc5 Merge pull request #8108 from radhikapc/building-qa
etcd/Documentation/dl_build.md: removed an extra step for testing etcd
2017-06-15 16:48:50 -07:00
Gyu-Ho Lee 5fedaf2dd7 Merge pull request #7896 from gyuho/metadata-grpc
*: gRPC v1.4.1, gateway v1.2.2, metadata Incoming/OutgoingContext
2017-06-15 16:42:55 -07:00
Gyu-Ho Lee 5e059fd8dc *: use metadata Incoming/OutgoingContext
Fix https://github.com/coreos/etcd/issues/7888.

Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 16:41:23 -07:00
Gyu-Ho Lee 0d0c0f3959 bill-of-materials: add google.golang.org/genproto
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 16:41:23 -07:00
Gyu-Ho Lee 5fe58228b4 vendor: update grpc-go v1.4.1, grpc-gateway v1.2.2
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 16:41:19 -07:00
Gyu-Ho Lee b9a53db0c2 Merge pull request #8101 from gyuho/randomize-renew
lease: randomize expiry on initial refresh call
2017-06-15 16:29:47 -07:00
Gyu-Ho Lee 639687bb89 Merge pull request #8112 from gyuho/speakeasy-dep
vendor: use tagged release 'bgentry/speakeasy'
2017-06-15 16:10:21 -07:00
Gyu-Ho Lee 15b86d064d vendor: use tagged release 'bgentry/speakeasy'
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 16:09:46 -07:00
Anthony Romano b6b56160cd Merge pull request #8111 from heyitsanthony/version-probing
vendor: update glide.yaml to use probing 0.0.1
2017-06-15 16:02:09 -07:00
Anthony Romano 703893f334 Merge pull request #8109 from heyitsanthony/faq-initial-peers
Documentation: update FAQ with entry about changing peer advertising
2017-06-15 16:01:19 -07:00
Anthony Romano 099952136a Merge pull request #8107 from heyitsanthony/lock-faster
concurrency: fetch current lock holder when creating waitlist key
2017-06-15 15:12:08 -07:00
Radhika Puthiyetath 52afc03d68 Documentation: removed an extra step for testing etcd
removed an extra step for testing etce build that might confuse the user of the flow. minimal editing to the doc
2017-06-15 14:39:10 -07:00
Anthony Romano 6e74c335e2 vendor: update glide.yaml to use probing 0.0.1
Also ignores appengine import from the grpc-gateway examples which
were causing glide errors on x/crypto when fetching imports.
2017-06-15 14:22:20 -07:00
Anthony Romano aa0e6b26c0 etcdserver: use RecorderStream for TestSyncTimeout to avoid missing action 2017-06-15 13:43:53 -07:00
Anthony Romano 44422f3898 Documentation: update FAQ with entry about changing peer advertising
Been seeing this somewhat frequently.
2017-06-15 13:31:25 -07:00
Anthony Romano dcf52bbfac etcdserver, embed, integration: don't use pointer for ServerConfig
ServerConfig is owned by etdcserver and unshared, so don't pass or store by
pointer. Also removes duplicated field 'snapCount'.
2017-06-15 13:02:13 -07:00
Gyu-Ho Lee 95bc33f37f integration: remove lease exist checking on randomized expiry
Lease with TTL 5 should be renewed with randomization,
thus it's still possible to exist after 3 seconds.

Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 12:57:55 -07:00
Gyu-Ho Lee 5bba05703c lease: randomize expiry on initial refresh call
Randomize the very first expiry on lease recovery
to prevent recovered leases from expiring all at
the same time.

Address https://github.com/coreos/etcd/issues/8096.

Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 12:57:49 -07:00
Gyu-Ho Lee 037e33e833 Merge pull request #8093 from gyuho/grafana
Documentation/op-guide: fix failed RPC rate, leader election metrics
2017-06-15 11:59:03 -07:00
Gyu-Ho Lee 1748fe3eda Documentation/op-guide: fix failed RPC rate, leader election metrics
This fixes failed RPC rate query, where we do not need
subtraction because we already query by the status code.
Also adds grpc_method to make it more specific. Most of the
time, the failure recovers within 10-second, which is our
Prometheus scrap interval, so 'rate' query might not cover
that time window, showing as 0s, but still shows up in the graph.

Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-06-15 11:54:18 -07:00
Anthony Romano f5b96991a1 concurrency: fetch current lock holder when creating waitlist key
The uncontended path for a mutex would fetch the minimum
revision key on the prefix after creating its entry in
the wait list. This fetch can be rolled into the txn for
creating the wait key, eliminating a round-trip for immediately
acquiring the lock.
2017-06-15 11:29:34 -07:00
Anthony Romano 1f206c027a Merge pull request #8106 from heyitsanthony/clarify-watch-comment
clientv3: clarify Watch close conditions
2017-06-15 10:56:25 -07:00
Xiang Li 3a37b68cda Merge pull request #8105 from nkovacs/its
Documentation: grammar fixes, it's -> its
2017-06-15 10:46:20 -07:00
Sam Batschelet c27634c215 e2e: test auth over grpc json 2017-06-15 13:41:47 -04:00
Sam Batschelet e5aa938fec scripts: generate swagger with authorization support 2017-06-15 13:41:43 -04:00
Anthony Romano 13d9438cf9 clientv3: clarify Watch close conditions
The "too slow" comment is rather vague. If the server closes
the watch for being too slow (it doesn't seem to any more), the
watch client should gracefully resume instead of forcing the
user to handle it.

Also removed the 'opts' comment since it wasn't being maintained.
2017-06-15 09:34:00 -07:00