前段时间，一套2节点rac 中的第一个节点数据库出现大量gc block lost ，业务系统也出现卡顿的现象，故障时段awr 如下:

       通过分析心跳网络延迟、网络丢包状况都正常，服务器硬件状况也是正常。

       通过netstat -s 指令进一步排查，发现大量的数据包重组失败错误(reassembles failed)

Ip:
209931138642 total packets received
13 forwarded
0 incoming packets discarded
159270412667 incoming packets delivered
126475270543 requests sent out
4498 outgoing packets dropped
3 dropped because of missing route
1460 fragments dropped after timeout
61955447119 reassemblies required
11294728390 packets reassembled ok
7343823 packet reassembles failed
5898092091 fragments received ok
26675594738 fragments created

参照oracle mos ，增加数据包重组缓冲区大小、数据包重组时间内核参数后恢复正常：

net.ipv4.ipfrag_high_thresh = 41943040
net.ipv4.ipfrag_low_thresh = 40894464
net.ipv4.ipfrag_time = 120