机器丢包了.

树袋熊

树袋熊
$ while true
> do
> netstat -s -p ip | grep "packets dropped due to the full socket receive buffer";
> sleep 10 ;
> done
        114661 packets dropped due to the full socket receive buffer
        114662 packets dropped due to the full socket receive buffer
        114663 packets dropped due to the full socket receive buffer
        114664 packets dropped due to the full socket receive buffer
        114665 packets dropped due to the full socket receive buffer

CLOSE_WAIT不多, 只有3行:

$ netstat -Aan | grep -i close
f1000700010493b0 tcp        0      0  *.*                *.*                CLOSED
f10007000104abb0 tcp4       0      0  *.*                *.*                CLOSED
f100070001f5ebb0 tcp        0      0  *.*                *.*                CLOSED
f100070001f40bb0 tcp4       0      0  *.*                *.*                CLOSED
f1000700026d33b0 tcp        0      0  *.*                *.*                CLOSED
f1000700026d1bb0 tcp4       0      0  *.*                *.*                CLOSED
f100070001e023b0 tcp        0      0  *.*                *.*                CLOSED
f100070001e02bb0 tcp4       0      0  *.*                *.*                CLOSED
f1000700006c63b0 tcp4      23      0  141.122.167.16.354 141.122.167.214.31 CLOSE_WAIT
f100070000961bb0 tcp4       0      0  141.122.167.16.949 141.122.168.108.21 CLOSE_WAIT
f1000700003c73b0 tcp4       0      0  141.122.166.240.94 141.122.168.108.23 CLOSE_WAIT


错误日志:

IDENTIFIER TIMESTAMP  T C RESOURCE_NAME  DESCRIPTION
E87EF1BE   0824150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0823150010 P O dumpcheck      The largest dump device is too small.
CB4A951F   0823024210 I S SRC            SOFTWARE PROGRAM ERROR
E87EF1BE   0822150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0821150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0820150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0819150010 P O dumpcheck      The largest dump device is too small.
5A7598C3   0819042710 I O fscsi9         Additional FC SCSI Protocol Driver Infor
5A7598C3   0819040710 I O fscsi9         Additional FC SCSI Protocol Driver Infor
E87EF1BE   0818150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0817150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0816150010 P O dumpcheck      The largest dump device is too small.
8650BE3F   0816013610 I H ent3           ETHERCHANNEL RECOVERY
F3846E13   0816013510 P H ent3           ETHERCHANNEL FAILOVER
E87EF1BE   0815150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0814150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0813150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0812150010 P O dumpcheck      The largest dump device is too small.
F3931284   0811230210 I H ent0           ETHERNET NETWORK RECOVERY MODE
F3931284   0811230110 I H ent4           ETHERNET NETWORK RECOVERY MODE
173C787F   0811225810 I S topsvcs        Possible malfunction on local adapter
EC0BCCD4   0811225810 T H ent0           ETHERNET DOWN
EC0BCCD4   0811225810 T H ent4           ETHERNET DOWN
E87EF1BE   0811150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0810150010 P O dumpcheck      The largest dump device is too small.
5FC2DD4B   0809175110 I H ent3           PING TO REMOTE HOST FAILED
E87EF1BE   0809150010 P O dumpcheck      The largest dump device is too small.
E87EF1BE   0808150010 P O dumpcheck      The largest dump device is too small.


机器性能有点影响,业务运转正常.
问题跟进中.
我们组另外一个人在跟.
我是有空就关注一下.

XQKA

>while true
> do
> netstat -s -p ip | grep "packets dropped due to the full socket receive buffer"
> sleep 10
> done
        3222 packets dropped due to the full socket receive buffer
        3243 packets dropped due to the full socket receive buffer
        3263 packets dropped due to the full socket receive buffer
        3283 packets dropped due to the full socket receive buffer
        3303 packets dropped due to the full socket receive buffer
        3323 packets dropped due to the full socket receive buffer
        3343 packets dropped due to the full socket receive buffer
        3364 packets dropped due to the full socket receive buffer
        3384 packets dropped due to the full socket receive buffer

是不是说明我的主机一直有丢包现象?

ivovoto

好样的，支持袋熊当版主!!!!

staryang01_cu

基本上就是 应用有问题--》数据库----》OS------》网络，
相互打太极。

树袋熊

[quote]谢谢，板主，速度真快，赞一个；
谢谢，板主，速度真快，赞一个；
$ while true
> do
> netstat -s -p ip | grep "packets dropped due to the full socket receive buffer";
> sleep 10 ;
> done
可否讲解下这几条命令，以前没用过，也没见过这样使用的，在机器上试了下，感觉真不错。




    while true
do
...
done

是循环.
sleep 10 是睡10秒再干.
netstat -s -p--> 这个就是干了, 但是因为是循环就睡10秒干一次活......
具体解释看MAN NETSTAT 就行了.

yddll

傻傻的问一句：
ps -ef |grep nim
hats_nim  & hats_diskhb_nim 进程.
然后杀掉hats_nim.

上面两个进 ...
abit2007 发表于 2010-09-15 10:58

    rsct的程序，hacmp有些东西是基于rsct来做的

abit2007

谢谢，板主，速度真快，赞一个；
$ while true
> do
> netstat -s -p ip | grep "packets dropped due to the full socket receive buffer";
> sleep 10 ;
> done
可否讲解下这几条命令，以前没用过，也没见过这样使用的，在机器上试了下，感觉真不错。

树袋熊

傻傻的问一句：
ps -ef |grep nim
hats_nim  & hats_diskhb_nim 进程.
然后杀掉hats_nim.

上面两个进 ...
abit2007 发表于 2010-09-15 10:58

    是的.和HA有关的,不是系统自己带的,杀完以后他会自动RESPAWN的.

abit2007

傻傻的问一句：
ps -ef |grep nim
hats_nim  & hats_diskhb_nim 进程.
然后杀掉hats_nim.

上面两个进程是HA的吗？

树袋熊

hacmp的bug？

网络异常导致心跳和写日志的异常？写日志是两节点都写的，写的多了，可能反过来 ...
yddll 发表于 2010-09-14 14:04

    也可以这么说.
    对,两节点都写.
    说是BUG吧,人家平时也运行挺好的,只能说错误自适应能力不够强.