Logstash——multiline 插件，匹配多行日志
本文内容
- 测试数据
- 字段属性
- 按多行解析运行时日志
- 把多行日志解析到字段
- 参考资料
在处理日志时，除了访问日志外，还要处理运行时日志，该日志大都用程序写的，比如 log4j。运行时日志跟访问日志最大的不同是，运行时日志是多行，也就是说，连续的多行才能表达一个意思。
本文主要说明，如何用 multiline 出来运行日志。代码如下（必须在filter中，否则结果还是不能合并行）：
filter {multiline { )}如果能按多行处理，那么把他们拆分到字段就很容易了。
测试数据
[16-04-12 03:40:01 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.
[16-04-12 03:40:02 DEBUG] impl.JdbcEntityInserter:- from product_category product_category
where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null
order by product_category.ORDERS asc
[16-04-12 03:40:03 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.
[16-04-12 03:40:04 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.
[16-04-12 03:40:05 DEBUG] impl.JdbcEntityInserter:- from product_category product_category
where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null
order by product_category.ORDERS desc
[16-04-12 03:40:06 DEBUG] impl.JdbcEntityInserter:- from product_category product_category
where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null
order by product_category.ORDERS asc
[16-04-12 03:40:07 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.
测试是在7秒内发生的（当然是假数据）。可以看到，第二、五、六秒的日志是多行的，有条SQL语句。其他是单行的。
字段属性
对 multiline 插件来说，有三个设置比较重要：negate、pattern 和 what。
negate
类型是 boolean
默认为
false
否定正则表达式（如果没有匹配的话）。
pattern
必须设置
类型为 string
没有默认值
要匹配的正则表达式。
what
必须设置
可以为
previous 或next没有默认值
如果正则表达式匹配了，那么该事件是属于下一个或是前一个事件？
按多行解析运行时日志
- sincedb_path
如果你不想用默认的 $HOME/.sincedb(Windows 平台上在C:\Windows\System32\config\systemprofile\.sincedb)，可以通过这个配置定义 sincedb 文件到其他位置。
- sincedb_write_interval
logstash 每隔多久写一次 sincedb 文件，默认是 15 秒。
- stat_interval
logstash 每隔多久检查一次被监听文件状态（是否有更新），默认是 1 秒。
- start_position
logstash 从什么位置开始读取文件数据，默认是结束位置，也就是说 logstash 进程会以类似 tail -F 的形式运行。如果你是要导入原有数据，把这个设定改成 "beginning"，logstash 进程就从头开始读取，有点类似cat，但是读到最后一行不会终止，而是继续变成 tail -F。
示例1：若配置文件如下所示，
input {
file{
path=>"/usr/local/elk/logstash/logs/c.out"
type=>"runtimelog"
codec=> multiline {
pattern => "^\["
negate => true
what => "previous"
}
start_position=>"beginning"
sincedb_path=>"/usr/local/elk/logstash/sincedb-access"
ignore_older=>0
}
}
output{
stdout{
codec=>rubydebug
}
}
说明：匹配以“[”开头的行，如果不是，那肯定是属于前一行的。
解析结果如下所示，能解析出6个JSON：
{
"@timestamp" => "2016-06-01T04:37:43.147Z",
"message" => "[16-04-12 03:40:01 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:37:43.152Z",
"message" => "[16-04-12 03:40:02 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:37:43.152Z",
"message" => "[16-04-12 03:40:03 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:37:43.155Z",
"message" => "[16-04-12 03:40:04 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:37:43.157Z",
"message" => "[16-04-12 03:40:05 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS desc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:37:43.159Z",
"message" => "[16-04-12 03:40:06 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
解析时，最后一行日志，不会解析。只有当再追加一条日志时，才会解析最后一条日志。
示例2：若将配置文件修改为，
input {
file{
path=>"/usr/local/elk/logstash/logs/c.out"
type=>"runtimelog"
codec=>multiline {
pattern => "^\["
negate => true
what => "next"
}
start_position=>"beginning"
sincedb_path=>"/usr/local/elk/logstash/sincedb-access"
ignore_older=>0
}
}
output{
stdout{
codec=>rubydebug
}
}
解析结果为，能解析出7个JSON：
{
"@timestamp" => "2016-06-01T04:40:43.232Z",
"message" => "[16-04-12 03:40:01 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.237Z",
"message" => "[16-04-12 03:40:02 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.238Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc\n[16-04-12 03:40:03 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.239Z",
"message" => "[16-04-12 03:40:04 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.244Z",
"message" => "[16-04-12 03:40:05 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.245Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS desc\n[16-04-12 03:40:06 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T04:40:43.249Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc\n[16-04-12 03:40:07 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
示例3：若将配置文件修改为，
codec=>multiline {
pattern => "^\["
negate => false
what => "previous"
}
则解析结果为：
{
"@timestamp" => "2016-06-01T05:38:50.853Z",
"message" => "[16-04-12 03:40:01 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.\n[16-04-12 03:40:02 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T05:38:50.856Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T05:38:50.858Z",
"message" => "order by product_category.ORDERS asc\n[16-04-12 03:40:03 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.\n[16-04-12 03:40:04 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.\n[16-04-12 03:40:05 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T05:38:50.860Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T05:38:50.861Z",
"message" => "order by product_category.ORDERS desc\n[16-04-12 03:40:06 DEBUG] impl.JdbcEntityInserter:- from product_category product_category",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
{
"@timestamp" => "2016-06-01T05:38:50.863Z",
"message" => "where product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog"
}
把多行日志解析到字段
配置文件如下所示：
input {
file{
path=>"/usr/local/elk/logstash/logs/c.out"
type=>"runtimelog"
codec=>multiline {
pattern => "^\["
negate => true
what => "previous"
}
start_position=>"beginning"
sincedb_path=>"/usr/local/elk/logstash/sincedb-access"
ignore_older=>0
}
}
filter {
multiline { pattern => "^\[" negate => true what => "previous" } date { match => [ "timestamp", "yyyy-MM-dd-HH:mm:ss" ] locale => "cn" }
grok {
match=>["message","\[%{TIMESTAMP_ISO8601:timestamp} %{LOGLEVEL:level}\] %{GREEDYDATA:msg}"]
}
}
output{
stdout{
codec=>rubydebug
}
}
解析后结果：
{
"@timestamp" => "2016-06-01T06:33:26.426Z",
"message" => "[16-04-12 03:40:01 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:01",
"level" => "DEBUG",
"msg" => "model.MappingNode:- ['/store/shopclass'] matched over."
}
{
"@timestamp" => "2016-06-01T06:33:26.485Z",
"message" => "[16-04-12 03:40:02 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:02",
"level" => "DEBUG",
"msg" => "impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc"
}
{
"@timestamp" => "2016-06-01T06:33:26.491Z",
"message" => "[16-04-12 03:40:03 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:03",
"level" => "DEBUG",
"msg" => "model.MappingNode:- ['/store/shopclass'] matched over."
}
{
"@timestamp" => "2016-06-01T06:33:26.492Z",
"message" => "[16-04-12 03:40:04 DEBUG] model.MappingNode:- ['/store/shopclass'] matched over.",
"@version" => "1",
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:04",
"level" => "DEBUG",
"msg" => "model.MappingNode:- ['/store/shopclass'] matched over."
}
{
"@timestamp" => "2016-06-01T06:33:26.494Z",
"message" => "[16-04-12 03:40:05 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS desc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:05",
"level" => "DEBUG",
"msg" => "impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS desc"
}
{
"@timestamp" => "2016-06-01T06:33:26.495Z",
"message" => "[16-04-12 03:40:06 DEBUG] impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc",
"@version" => "1",
"tags" => [
[0] "multiline"
],
"path" => "/usr/local/elk/logstash/logs/c.out",
"host" => "vcyber",
"type" => "runtimelog",
"timestamp" => "16-04-12 03:40:06",
"level" => "DEBUG",
"msg" => "impl.JdbcEntityInserter:- from product_category product_category\nwhere product_category.PARENT_ID is null and product_category.STATUS = ? and product_category.DEALER_ID is null\norder by product_category.ORDERS asc"
}
参考资料
- Logstash multiline
转载自：http://www.cnblogs.com/liuning8023/archive/2016/06/01/5549366.html