如何解决Logstash多行无法正确解析Oracle日志
我一直在使用Logstash管道进行测试，以处理具有以下格式的Oracle多行审核日志：
Audit file /u01/app/oracle/admin/DEVINST/adump/DEVINST_ora_15460_20201001230100743853143795.aud
Oracle Database 12c Standard Edition Release 12.2.0.1.0 - 64bit DEV
Build label: RDBMS_12.2.0.1.0_LINUX.X64_170125
ORACLE_HOME: /u01/app/oracle/product/12.2.0/dbhome_1
System name: Linux
Node name: testdevserver
Release: 3.10.0-862.14.4.el7.x86_64
Version: #1 SMP Fri Sep 21 09:07:21 UTC 2018
Machine: x86_64
Instance name: DEVINST
Redo thread mounted by this instance: 1
Oracle process number: 57
Unix process pid: 15460,image: oracle@testdevserver (TNS V1-V3)
Thu Oct 1 23:01:00 2020 +00:00
LENGTH : '275'
ACTION :[7] 'CONNECT'
DATABASE USER:[1] '/'
PRIVILEGE :[6] 'SYSDBA'
CLIENT USER:[9] 'test_user'
CLIENT TERMINAL:[5] 'pts/0'
STATUS:[1] '0'
DBID:[10] '1762369616'
SESSIONID:[10] '4294967295'
USERHOST:[21] 'testdevserver'
CLIENT ADDRESS:[0] ''
ACTION NUMBER:[3] '100'
Thu Oct 1 23:01:00 2020 +00:00
LENGTH : '296'
ACTION :[29] 'SELECT STATUS FROM V$INSTANCE'
DATABASE USER:[1] '/'
PRIVILEGE :[6] 'SYSDBA'
CLIENT USER:[9] 'test_user'
CLIENT TERMINAL:[5] 'pts/0'
STATUS:[1] '0'
DBID:[10] '1762369616'
SESSIONID:[10] '4294967295'
USERHOST:[21] 'testdevserver'
CLIENT ADDRESS:[0] ''
ACTION NUMBER:[1] '3'
我的/etc/logstash/conf.d/25-filter.conf：
filter {
grok {
match => { "message" => "(?<day>^[A-Za-z]{3}) (?<month>[A-Za-z]{3}) (?<monthday>[0-9]{2}) (?<hour>[0-9]{2}):(?<min>[0-9]{2}):(?<sec>[0-9]{2}) (?<year>[0-9]{4}) %{GREEDYDATA:message}" }
overwrite => [ "message" ]
add_tag => [ "oracle_audit" ]
}
grok {
match => { "ACTION :\[[0-9]*\] '(?<ora_audit_action>.*)'.*DATABASE USER:\[[0-9]*\] '(?<ora_audit_dbuser>.*)'.*PRIVILEGE :\[[0-9]*\] '(?<ora_audit_priv>.*)'.*CLIENT USER:\[[0-9]*\] '(?<ora_audit_osuser>.*)'.*CLIENT TERMINAL:\[[0-9]*\] '(?<ora_audit_term>.*)'.*STATUS:\[[0-9]*\] '(?<ora_audit_status>.*)'.*DBID:\[[0-9]*\] '(?<ora_audit_dbid>.*)'.*SESSIONID:\[[0-9]*\] '(?<ora_audit_sessionid>.*)'.*USERHOST:\[[0-9]*\] '(?<ora_audit_dbhost>.*)'.*CLIENT ADDRESS:\[[0-9]*\] '(?<ora_audit_clientaddr>.*)'.*ACTION NUMBER:\[[0-9]*\] '(?<ora_audit_actionnum>.*)'" }
}
grok {
match => { "source" => [ ".*/[a-zA-Z0-9_#$]*_[a-z0-9]*_(?<ora_audit_derived_pid>[0-9]*)_[0-9]*\.aud" ] }
}
mutate {
add_field => { "timestamp" => "%{year}-%{day}-%{monthday} %{hour}:%{min}:%{sec}" }
}
date {
locale => "en"
match => [ "timestamp","YYYY-MMM-dd HH:mm:ss" ]
}
mutate {
remove_field => [ "timestamp","year","day","monthday","hour","min","sec" ]
}
}
我的/etc/logstash/conf.d/000-file-in.conf文件：
input {
file {
path => [ "/tmp/testora" ]
start_position => "beginning"
codec => multiline
{
pattern => "^[A-Za-z]{3} [A-Za-z]{3} [0-9]{2} [0-9]{2}:[0-9]{2}:[0-9]{2} [0-9]{4}"
negate => "true"
what => "previous"
}
}
}
然后我通过运行进行测试：
/usr/share/logstash/bin/logstash -r -f /etc/logstash/conf.d/：
....
[INFO ] 2020-10-05 11:16:30.656 [Converge PipelineAction::Reload<main>] reload - Reloading pipeline {"pipeline.id"=>:main}
[INFO ] 2020-10-05 11:16:30.662 [Converge PipelineAction::Reload<main>] observingtail - QUIT - closing all files and shutting down.
{
"ora_audit_dbid" => "1762369616","ora_audit_actionnum" => "3","ora_audit_sessionid" => "4294967295","@version" => "1","tags" => [
[0] "multiline",[1] "_grokparsefailure",[2] "_dateparsefailure"
],"path" => "/tmp/testora","ora_audit_action" => [
[0] "275'\nACTION :[7] 'CONNECT'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[3] '100'\nThu Oct 1 23:01:00 2020 +00:00\nLENGTH : '296",[1] "SELECT STATUS FROM V$INSTANCE"
],"@timestamp" => 2020-10-05T01:16:30.889Z,"ora_audit_priv" => "SYSDBA","message" => "Audit file /u01/app/oracle/admin/DEVINST/adump/DEVINST_ora_15460_20201001230100743853143795.aud\nOracle Database 12c Standard Edition Release 12.2.0.1.0 - 64bit DEV\nBuild label: RDBMS_12.2.0.1.0_LINUX.X64_170125\nORACLE_HOME: /u01/app/oracle/product/12.2.0/dbhome_1\nSystem name: Linux\nNode name: testdevserver\nRelease: 3.10.0-862.14.4.el7.x86_64\nVersion: #1 SMP Fri Sep 21 09:07:21 UTC 2018\nMachine: x86_64\nInstance name: DEVINST\nRedo thread mounted by this instance: 1\nOracle process number: 57\nUnix process pid: 15460,image: oracle@testdevserver (TNS V1-V3)\nThu Oct 1 23:01:00 2020 +00:00\nLENGTH : '275'\nACTION :[7] 'CONNECT'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[3] '100'\nThu Oct 1 23:01:00 2020 +00:00\nLENGTH : '296'\nACTION :[29] 'SELECT STATUS FROM V$INSTANCE'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[1] '3'","ora_audit_dbhost" => "testdevserver","host" => "myhost","ora_audit_dbuser" => "/","ora_audit_osuser" => "test_user","ora_audit_term" => "pts/0","ora_audit_status" => "0"
}
不幸的是，这不是我所期望的。不知何故，它没有对消息进行分块和正确地解析消息。我期待的是类似的东西：
"ora_audit_action" => "CONNECT","ora_audit_dbid" => "1762369616","ora_audit_status" => "0","ora_audit_clientaddr" => "","message" => "Thu Oct 1 23:01:00 2020 +00:00\nLENGTH : '275'\nACTION :[7] 'CONNECT'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[3] '100'\nThu Oct 1 23:01:00 2020 +00:00\nLENGTH : '296'\nACTION :[29] 'SELECT STATUS FROM V$INSTANCE'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[1] '3'"
我看到了其中一些信息，但是我也希望它能吐出2条“大块”消息（与模式^[A-Za-z]{3} [A-Za-z]{3} [0-9]{2} [0-9]{2}:[0-9]{2}:[0-9]{2} [0-9]{4}相匹配，作为行的开头，例如Thu Oct 1 23:01:00 2020 +00:00） -相反，它看起来只看到一个？我认为我的模式匹配可能是这里的问题，如果有人可以提供任何提示，谢谢。
此外，我不确定是什么导致了这些错误[1] "_grokparsefailure",[2] "_dateparsefailure"-显然，它无法正确解析内容，但我只是不知道该怎么做。
帮助：（
感谢J
解决方法
在与Thu Oct 1 23:01:00 2020 +00:00 匹配的Grok模式下找到
(?<timestamp>%{DAY} %{MONTH} %{MONTHDAY} %{TIME})
输出的屏幕截图：
我认为以下模式应适用于您的多行编解码器并提供所需的分组。
pattern => "^([A-Za-z]{3})(\s*)([A-Za-z]{3})(\s*)([0-9]{1,2})(\s*)([0-9]{2}:[0-9]{2}:[0-9]{2} [0-9]{4})"
此外，在grok过滤器下方，应能为您提供多线的所需细目分类。您可能想根据喜好重命名密钥。
input {
file {
path => [ "/tmp/testora" ]
start_position => "beginning"
codec => multiline
{
pattern => "^([A-Za-z]{3})(\s*)([A-Za-z]{3})(\s*)([0-9]{1,2})(\s*)([0-9]{2}:[0-9]{2}:[0-9]{2} [0-9]{4})"
negate => "true"
what => "previous"
}
}
}
filter {
grok {
match => { "message" => "(?<day>[A-Za-z]{3})%{SPACE}(?<month>[A-Za-z]{3})%{SPACE}(?<monthday>[0-9]{1,2})%{SPACE}(?<hour>[0-9]{1,2}):(?<min>[0-9]{1,2}):(?<sec>[0-9]{2})%{SPACE}(?<year>[0-9]{4})%{SPACE}%{GREEDYDATA:message}" }
overwrite => [ "message" ]
add_tag => [ "oracle_audit" ]
}
kv {
field_split => "\n"
value_split_pattern => "\s*:\s*\[[0-9]*\]\s*"
source => "message"
}
mutate {
add_field => { "timestamp" => "%{year}-%{month}-%{monthday} %{hour}:%{min}:%{sec}" }
}
date {
locale => "en"
match => [ "timestamp","YYYY-MMM-dd HH:mm:ss" ]
}
mutate {
remove_field => [ "timestamp","year","day","monthday","month","hour","min","sec" ]
}
}
output {
stdout {}
}
管道的示例输出如下
{
"PRIVILEGE" => "SYSDBA","DBID" => "1762369616","ACTION NUMBER" => "100","USERHOST" => "testdevserver","@version" => "1","path" => "/usr/share/logstash/stack/data/file.txt","SESSIONID" => "4294967295","CLIENT USER" => "test_user","CLIENT TERMINAL" => "pts/0","STATUS" => "0","host" => "e8f97b3e53ff","tags" => [
[0] "multiline",[1] "oracle_audit"
],"ACTION" => "CONNECT","DATABASE USER" => "/","@timestamp" => 2020-10-01T23:01:00.000Z,"message" => "+00:00\nLENGTH : '275'\nACTION :[7] 'CONNECT'\nDATABASE USER:[1] '/'\nPRIVILEGE :[6] 'SYSDBA'\nCLIENT USER:[9] 'test_user'\nCLIENT TERMINAL:[5] 'pts/0'\nSTATUS:[1] '0'\nDBID:[10] '1762369616'\nSESSIONID:[10] '4294967295'\nUSERHOST:[21] 'testdevserver'\nCLIENT ADDRESS:[0] ''\nACTION NUMBER:[3] '100'"
}
根据您的要求，您必须修改希腊模式 在第二个针对LENGH和ACTION参数的grok过滤器中，您正在创建相同的字段 ora_audit_action ，因此它将对数据进行打包并形成一个数组，以便为两个参数分别创建字段
grok {
match => { "msg1" => "LENGTH : '(?<ora_audit_length>.*)'.ACTION :\[[0-9]*\] '(?<ora_audit_action>.*)'.*DATABASE USER:\[[0-9]*\] '(?<ora_audit_dbuser>.*)'.*PRIVILEGE :\[[0-9]*\] '(?<ora_audit_priv>.*)'.*CLIENT USER:\[[0-9]*\] '(?<ora_audit_osuser>.*)'.*CLIENT TERMINAL:\[[0-9]*\] '(?<ora_audit_term>.*)'.*STATUS:\[[0-9]*\] '(?<ora_audit_status>.*)'.*DBID:\[[0-9]*\] '(?<ora_audit_dbid>.*)'.*SESSIONID:\[[0-9]*\] '(?<ora_audit_sessionid>.*)'.*USERHOST:\[[0-9]*\] '(?<ora_audit_dbhost>.*)'.*CLIENT ADDRESS:\[[0-9]*\] '(?<ora_audit_clientaddr>.*)'.*ACTION NUMBER:\[[0-9]*\] '(?<ora_audit_actionnum>.*)'" }
}
代替删除字段，您可以使用修剪过滤器将必要的字段列入白名单。
prune {
whitelist_names => ["ora_audit_action","ora_audit_dbuser","ora_audit_dbid","ora_audit_status","ora_audit_osuser","ora_audit_priv","ora_audit_term","ora_audit_sessionid","ora_audit_dbhost","ora_audit_clientaddr","ora_audit_actionnum","host","@timestamp","@version","message"]
}
在代码中，您将隔离时间戳并对其进行重整，但在过滤器部分中，您将其删除，并使用默认的 @timestamp 字段，如果您没有要求，则可以删除不必要的代码。
版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 dio@foxmail.com 举报，一经查实，本站将立刻删除。