Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add periodic health checks to TChannel (#318)
If a TChannel connection stalls (e.g., the remote side has disappeared, and traffic starts black-holing), TChannel waits for TCP to detect this issue, but this is often not configurable, and it can be a long period of time (we've observed many minutes in production). In many P2P situations, it's desirable to detect these failures as soon as possible. Support active health checking using TChannel pings on a periodic basis, with configurable timeouts, number of failures etc.
- Loading branch information
Showing
8 changed files
with
568 additions
and
19 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,168 @@ | ||
| // Copyright (c) 2017 Uber Technologies, Inc. | ||
|
|
||
| // Permission is hereby granted, free of charge, to any person obtaining a copy | ||
| // of this software and associated documentation files (the "Software"), to deal | ||
| // in the Software without restriction, including without limitation the rights | ||
| // to use, copy, modify, merge, publish, distribute, sublicense, and/or sell | ||
| // copies of the Software, and to permit persons to whom the Software is | ||
| // furnished to do so, subject to the following conditions: | ||
| // | ||
| // The above copyright notice and this permission notice shall be included in | ||
| // all copies or substantial portions of the Software. | ||
| // | ||
| // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR | ||
| // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, | ||
| // FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE | ||
| // AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER | ||
| // LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, | ||
| // OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN | ||
| // THE SOFTWARE. | ||
|
|
||
| package tchannel | ||
|
|
||
| import ( | ||
| "sync" | ||
| "time" | ||
|
|
||
| "golang.org/x/net/context" | ||
| ) | ||
|
|
||
| const ( | ||
| _defaultHealthCheckTimeout = time.Second | ||
| _defaultHealthCheckFailuresToClose = 5 | ||
|
|
||
| _healthHistorySize = 256 | ||
| ) | ||
|
|
||
| // HealthCheckOptions are the parameters to configure active TChannel health | ||
| // checks. These are not intended to check application level health, but | ||
| // TCP connection health (similar to TCP keep-alives). The health checks use | ||
| // TChannel ping messages. | ||
| type HealthCheckOptions struct { | ||
| // The period between health checks. If this is zeor, active health checks | ||
| // are disabled. | ||
| Interval time.Duration | ||
|
|
||
| // The timeout to use for a health check. | ||
| // If no value is specified, it defaults to time.Second. | ||
| Timeout time.Duration | ||
|
|
||
| // FailuresToClose is the number of consecutive health check failures that | ||
| // will cause this connection to be closed. | ||
| // If no value is specified, it defaults to 5. | ||
| FailuresToClose int | ||
| } | ||
|
|
||
| type healthHistory struct { | ||
| sync.RWMutex | ||
|
|
||
| states []bool | ||
|
|
||
| insertAt int | ||
| total int | ||
| } | ||
|
|
||
| func newHealthHistory() *healthHistory { | ||
| return &healthHistory{ | ||
| states: make([]bool, _healthHistorySize), | ||
| } | ||
| } | ||
|
|
||
| func (hh *healthHistory) add(b bool) { | ||
| hh.Lock() | ||
| defer hh.Unlock() | ||
|
|
||
| hh.states[hh.insertAt] = b | ||
| hh.insertAt = (hh.insertAt + 1) % _healthHistorySize | ||
| hh.total++ | ||
| } | ||
|
|
||
| func (hh *healthHistory) asBools() []bool { | ||
| hh.RLock() | ||
| defer hh.RUnlock() | ||
|
|
||
| if hh.total < _healthHistorySize { | ||
| return append([]bool(nil), hh.states[:hh.total]...) | ||
| } | ||
|
|
||
| states := hh.states | ||
| copyStates := make([]bool, 0, _healthHistorySize) | ||
| copyStates = append(copyStates, states[hh.insertAt:]...) | ||
| copyStates = append(copyStates, states[:hh.insertAt]...) | ||
| return copyStates | ||
| } | ||
|
|
||
| func (hco HealthCheckOptions) enabled() bool { | ||
| return hco.Interval > 0 | ||
| } | ||
|
|
||
| func (hco HealthCheckOptions) withDefaults() HealthCheckOptions { | ||
| if hco.Timeout == 0 { | ||
| hco.Timeout = _defaultHealthCheckTimeout | ||
| } | ||
| if hco.FailuresToClose == 0 { | ||
| hco.FailuresToClose = _defaultHealthCheckFailuresToClose | ||
| } | ||
| return hco | ||
| } | ||
|
|
||
| // healthCheck will do periodic pings on the connection to check the state of the connection. | ||
| // We accept connID on the stack so can more easily debug panics or leaked goroutines. | ||
| func (c *Connection) healthCheck(connID uint32) { | ||
| opts := c.opts.HealthChecks | ||
|
|
||
| ticker := c.timeTicker(opts.Interval) | ||
| defer ticker.Stop() | ||
|
|
||
| consecutiveFailures := 0 | ||
| for { | ||
| select { | ||
| case <-ticker.C: | ||
| case <-c.healthCheckCtx.Done(): | ||
| return | ||
| } | ||
|
|
||
| ctx, cancel := context.WithTimeout(c.healthCheckCtx, opts.Timeout) | ||
| err := c.ping(ctx) | ||
| cancel() | ||
| c.healthCheckHistory.add(err == nil) | ||
| if err == nil { | ||
| if c.log.Enabled(LogLevelDebug) { | ||
| c.log.Debug("Performed successful active health check.") | ||
| } | ||
| consecutiveFailures = 0 | ||
| continue | ||
| } | ||
|
|
||
| // If the health check failed because the connection closed or health | ||
| // checks were stopped, we don't need to log or close the connection. | ||
| if GetSystemErrorCode(err) == ErrCodeCancelled || err == ErrInvalidConnectionState { | ||
| c.log.WithFields(ErrField(err)).Debug("Health checker stopped.") | ||
| return | ||
| } | ||
|
|
||
| consecutiveFailures++ | ||
| c.log.WithFields(LogFields{ | ||
| {"consecutiveFailures", consecutiveFailures}, | ||
| ErrField(err), | ||
| {"failuresToClose", opts.FailuresToClose}, | ||
| }...).Warn("Failed active health check.") | ||
|
|
||
| if consecutiveFailures >= opts.FailuresToClose { | ||
| c.close(LogFields{ | ||
| {"reason", "health check failure"}, | ||
| ErrField(err), | ||
| }...) | ||
| return | ||
| } | ||
| } | ||
| } | ||
|
|
||
| func (c *Connection) stopHealthCheck() { | ||
| // Best effort check to see if health checks were stopped. | ||
| if c.healthCheckCtx.Err() != nil { | ||
| return | ||
| } | ||
| c.log.Debug("Stopping health checks.") | ||
| c.healthCheckQuit() | ||
| } |
Oops, something went wrong.