我们在一台linux服务器上安装了hadoop cluster 2.6.4版metrics collector

root@master02 ambari-metrics-collector]#  rpm -qa | grep metrics-collector
ambari-metrics-collector-2.6.1.0-143.x86_64

在/var/log/ambari metrics collector下，我们有去年7月的gc.log文件，我们希望自动删除循环的gc.log文件

[root@master02 ambari-metrics-collector]# ls -ltr | grep gc.log | grep -v collector| more
-rw-r--r--. 1 ams  hadoop      1430 Jul 13  2017 gc.log-201707131628
-rw-r--r--  1 ams  hadoop 178518767 Aug 28  2017 gc.log-201707160637
-rw-r--r--  1 ams  hadoop 128748829 Sep 17  2017 gc.log-201708280731
-rw-r--r--  1 ams  hadoop     60849 Sep 18  2017 gc.log-201709181421
-rw-r--r--  1 ams  hadoop     82562 Sep 18  2017 gc.log-201709181424
-rw-r--r--  1 ams  hadoop 164381669 Oct 19  2017 gc.log-201709181508
-rw-r--r--  1 ams  hadoop     46149 Oct 19  2017 gc.log-201710191738
-rw-r--r--  1 ams  hadoop  13763796 Oct 23  2017 gc.log-201710200633
-rw-r--r--  1 ams  hadoop   1048289 Oct 23  2017 gc.log-201710231021
-rw-r--r--  1 ams  hadoop   3473874 Oct 24  2017 gc.log-201710231649
-rw-r--r--  1 ams  hadoop 137289030 Nov 15  2017 gc.log-201710241352
-rw-r--r--  1 ams  hadoop   8037548 Nov 16  2017 gc.log-201711151355
-rw-r--r--  1 ams  hadoop  23919164 Nov 19  2017 gc.log-201711161246
-rw-r--r--  1 ams  hadoop 139759594 Dec  6 16:35 gc.log-201711191241
-rw-r--r--  1 ams  hadoop 184118442 Dec 25 16:13 gc.log-201712061635
-rw-r--r--  1 ams  hadoop      1440 Jan  1 04:38 gc.log-201712251613
-rw-r--r--  1 ams  hadoop      1437 Jan 10 00:03 gc.log-201801021532
-rw-r--r--  1 ams  hadoop      1439 Jan 10 01:04 gc.log-201801100004
-rw-r--r--  1 ams  hadoop      3637 Jan 10 01:11 gc.log-201801100105
-rw-r--r--  1 ams  hadoop      1437 Jan 10 01:16 gc.log-201801100113
-rw-r--r--  1 ams  hadoop      3636 Jan 10 01:24 gc.log-201801100118
-rw-r--r--  1 ams  hadoop      1438 Jan 10 01:35 gc.log-201801100126
-rw-r--r--  1 ams  hadoop      1438 Jan 10 01:42 gc.log-201801100137
-rw-r--r--  1 ams  hadoop      1436 Jan 10 02:14 gc.log-201801100144
-rw-r--r--  1 ams  hadoop      8353 Jan 10 02:19 gc.log-201801100216
-rw-r--r--  1 ams  hadoop      1437 Jan 11 09:08 gc.log-201801110730
-rw-r--r--  1 ams  hadoop      1436 Jan 11 09:19 gc.log-201801110909
-rw-r--r--  1 ams  hadoop       766 Jan 11 09:20 gc.log-201801110920
-rw-r--r--  1 ams  hadoop       765 Jan 11 09:26 gc.log-201801110926

从ambari gui中的高级ams hbase env中，我们可以看到以下内容：

{% if java_version < 8 %}
export HBASE_MASTER_OPTS=" -XX:PermSize=64m -XX:MaxPermSize={{hbase_master_maxperm_size}} -Xms{{hbase_heapsize}} -Xmx{{hbase_heapsize}} -Xmn{{hbase_master_xmn_size}} -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly"
export HBASE_REGIONSERVER_OPTS="-XX:MaxPermSize=128m -Xmn{{regionserver_xmn_size}} -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -Xms{{regionserver_heapsize}} -Xmx{{regionserver_heapsize}}"
{% else %}
export HBASE_MASTER_OPTS=" -Xms{{hbase_heapsize}} -Xmx{{hbase_heapsize}} -Xmn{{hbase_master_xmn_size}} -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly"
export HBASE_REGIONSERVER_OPTS=" -Xmn{{regionserver_xmn_size}} -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -Xms{{regionserver_heapsize}} -Xmx{{regionserver_heapsize}}"
{% endif %}

我们尝试在hbase\u master\u opts或hbase\u regionserver\u opts中设置
为了只获取10个gc.log文件，-xx:numberofgclogfiles=10，但是metrics collector未能启动，似乎是因为语法不正确
参考-http://www.oracle.com/technetwork/java/vmoptions-jsp-140102.html

export HBASE_MASTER_OPTS=" -XX:PermSize=64m -XX:MaxPermSize={{hbase_master_maxperm_size}} -Xms{{hbase_heapsize}} -Xmx{{hbase_heapsize}} -Xmn{{hbase_master_xmn_size}} -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -XX:NumberOfGClogFiles=1"

请告知如何自动删除旋转的gc.log文件