From cf568487c8dc6d20a6e353e291d1f6184a200c08 Mon Sep 17 00:00:00 2001
From: Wu Taizeng <wutz@paratera.com>
Date: Tue, 26 Jan 2016 16:19:34 +0800
Subject: [PATCH] Fix some inputs panic will lead to the telegraf exit
closes #585
closes #584
---
CHANGELOG.md | 1 +
agent.go | 16 ++++++++++++++++
2 files changed, 17 insertions(+)
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 63611e3e..bc0f3fdd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -40,6 +40,7 @@ specifying a docker endpoint to get metrics from.
- [#440](https://github.com/influxdata/telegraf/issues/440): Don't query filtered devices for disk stats.
- [#463](https://github.com/influxdata/telegraf/issues/463): Docker plugin not working on AWS Linux
- [#568](https://github.com/influxdata/telegraf/issues/568): Multiple output race condition.
+- [#585](https://github.com/influxdata/telegraf/pull/585): Log stack trace and continue on Telegraf panic. Thanks @wutaizeng!
## v0.10.0 [2016-01-12]
diff --git a/agent.go b/agent.go
index d0f82145..ee5f4502 100644
--- a/agent.go
+++ b/agent.go
@@ -7,6 +7,7 @@ import (
"math/big"
"math/rand"
"os"
+ "runtime"
"sync"
"time"
@@ -87,6 +88,18 @@ func (a *Agent) Close() error {
return err
}
+func panicRecover(input *models.RunningInput) {
+ if err := recover(); err != nil {
+ trace := make([]byte, 2048)
+ runtime.Stack(trace, true)
+ log.Printf("FATAL: Input [%s] panicked: %s, Stack:\n%s\n",
+ input.Name, err, trace)
+ log.Println("PLEASE REPORT THIS PANIC ON GITHUB with " +
+ "stack trace, configuration, and OS information: " +
+ "https://github.com/influxdata/telegraf/issues/new")
+ }
+}
+
// gatherParallel runs the inputs that are using the same reporting interval
// as the telegraf agent.
func (a *Agent) gatherParallel(pointChan chan *client.Point) error {
@@ -103,6 +116,7 @@ func (a *Agent) gatherParallel(pointChan chan *client.Point) error {
wg.Add(1)
counter++
go func(input *models.RunningInput) {
+ defer panicRecover(input)
defer wg.Done()
acc := NewAccumulator(input.Config, pointChan)
@@ -148,6 +162,8 @@ func (a *Agent) gatherSeparate(
input *models.RunningInput,
pointChan chan *client.Point,
) error {
+ defer panicRecover(input)
+
ticker := time.NewTicker(input.Config.Interval)
for {
--
GitLab