Memcache Exception Timeout Diagnostic--mctop

Check the MC connection time out yesterday, found that the other MC is also a business did not find an exception, through the monitoring found that the network is not dropped packets, but the two MC traffic obviously inconsistent, in the case of the same number of connections, the bandwidth is 3 times times the difference, it can be concluded that a key value of the business call is too large.

Well, you know why. How to locate the key? The following are general practices:

1. Frequent request packets that are found by grabbing the packet;

2. The MC protocol is adopted to analyze whether the key value is too large;

It seems that you are a little dizzy, hehe, this method is very cumbersome, positioning reason is relatively slow, there is no better way? Below is the introduction of the next Mctop

Mctop is an open source detection tool that relies on Ruby's pcap for packet Capture analysis, which Githup https://github.com/etsy/mctop

Unzip execution under:

#./mctop

/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in ' Gem_original_require ': No such file to load--pcap ( Loaderror)

From/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in ' require '

From./.. /lib/cmdline.rb:2

From/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in ' Gem_original_require '

From/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in ' require '

From./mctop:10

Through the error learned, need to install PCAP, but found out there is still a problem, no way to upgrade Ruby Bar, I am now 1.8.7 upgrade to 1.9.2, you need to install the following components

#/usr/local/bin/gem Install b Undle

#/usr/local/bin /gem Install Rake

#/usr/local/bin/gem Install Ruby -pcap

#rm-F/usr/bin/ruby

#ln-S/usr/local/bin/ruby/usr/bin/ruby

Again run under

#./mctop--help

usage:mctop [Options]

-I,--interface=nic Network interface to sniff (required)

--host=hOST Network host to sniff in (default all)

-P,--port=port Network port to sniff on (default 11211)

-D,--discard=thresh discard keys with request/sec rate below Thresh

-R,--refresh=ms Refresh The stats display every MS milliseconds

-H,--help Show usage info

Then use it to monitor the abnormal ports:

#./mctop-i eth1--host=10.x.x.85-p Xxx-r 1000

Such as

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M00/58/4A/wKiom1St22ujm2YRAAEAIoCIv4Q407.jpg "title=" Mctop.jpg "alt=" Wkiom1st22ujm2yraaeaiociv4q407.jpg "/>

One eye to see, is esf-wordfilter__all value too large accounted for the 4m/s bandwidth, inform the Business Unit optimization of ~

Summarize:

Mctop is a very good online monitoring program, but also to inspire, you can do a set of MC monitoring according to this, well, that's it ~

This article is from the "Flying Snail" blog, please be sure to keep this source http://zhangbo.blog.51cto.com/350645/1600512

Memcache Exception Timeout Diagnostic--mctop