Mirroristas/etcd
2
0
Fork 0
mirror of https://github.com/etcd-io/etcd.git synced 2024-09-27 06:25:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

clientv3: clarify "WithRequireLeader" for network partition

Browse Source 
Signed-off-by: Gyuho Lee <gyuhox@gmail.com>
This commit is contained in:
Gyuho Lee 2018-06-20 10:28:41 -07:00
parent 897c055940
commit 11ead62b9d
1 changed files with 13 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
clientv3/watch.go
View File

@ -50,13 +50,19 @@ type Watcher interface {
// and "WatchResponse" from this closed channel has zero events and nil "Err()". // and "WatchResponse" from this closed channel has zero events and nil "Err()".
// The context "ctx" MUST be canceled, as soon as watcher is no longer being used, // The context "ctx" MUST be canceled, as soon as watcher is no longer being used,
// to release the associated resources. // to release the associated resources.
// If the context is "context.Background/TODO", returned "WatchChan" will not be closed //
// and wait until events happen, except when server returns a non-recoverable error. // If the context is "context.Background/TODO", returned "WatchChan" will
// For example, when context passed with "WithRequireLeader" and the connected server // not be closed and block until event is triggered, except when server
// has no leader, error "etcdserver: no leader" is returned, and then "WatchChan" is // returns a non-recoverable error (e.g. ErrCompacted).
// closed with non-nil "Err()". // For example, when context passed with "WithRequireLeader" and the
// Otherwise, as long as the context has not been canceled or timed out, watch will // connected server has no leader (e.g. due to network partition),
// retry on other recoverable errors forever until reconnected. // error "etcdserver: no leader" (ErrNoLeader) will be returned,
// and then "WatchChan" is closed with non-nil "Err()".
// In order to prevent a watch stream being stuck in a partitioned node,
// make sure to wrap context with "WithRequireLeader".
//
// Otherwise, as long as the context has not been canceled or timed out,
// watch will retry on other recoverable errors forever until reconnected.
// //
// TODO: explicitly set context error in the last "WatchResponse" message and close channel? // TODO: explicitly set context error in the last "WatchResponse" message and close channel?
// Currently, client contexts are overwritten with "valCtx" that never closes. // Currently, client contexts are overwritten with "valCtx" that never closes.