-
-
Save bkimble/1365005 to your computer and use it in GitHub Desktop.
#!/usr/bin/env ruby | |
# List all keys stored in memcache. | |
# Credit to Graham King at http://www.darkcoding.net/software/memcached-list-all-keys/ for the original article on how to get the data from memcache in the first place. | |
require 'net/telnet' | |
headings = %w(id expires bytes cache_key) | |
rows = [] | |
localhost = Net::Telnet::new("Host" => "localhost", "Port" => 11211, "Timeout" => 3) | |
matches = localhost.cmd("String" => "stats items", "Match" => /^END/).scan(/STAT items:(\d+):number (\d+)/) | |
slabs = matches.inject([]) { |items, item| items << Hash[*['id','items'].zip(item).flatten]; items } | |
longest_key_len = 0 | |
slabs.each do |slab| | |
localhost.cmd("String" => "stats cachedump #{slab['id']} #{slab['items']}", "Match" => /^END/) do |c| | |
matches = c.scan(/^ITEM (.+?) \[(\d+) b; (\d+) s\]$/).each do |key_data| | |
cache_key, bytes, expires_time = key_data | |
rows << [slab['id'], Time.at(expires_time.to_i), bytes, cache_key] | |
longest_key_len = [longest_key_len,cache_key.length].max | |
end | |
end | |
end | |
row_format = %Q(|%8s | %28s | %12s | %-#{longest_key_len}s |) | |
puts row_format%headings | |
rows.each{|row| puts row_format%row} | |
localhost.close |
As I didn't know about this snippet I created a similar one based on the same source. Please have a look if you find it useful.
require 'net/telnet'
memcached = Net::Telnet::new("Host" => "localhost", "Port" => 11211)
memcached.puts 'stats items'
slab_ids = memcached.waitfor(/./).split("\n").map { |line| x = line[/([^\:]+)\d+/, 0].to_i ; x > 0 ? x : nil }.uniq.compact
slab_ids.inject([]) do |items, slab_id|
memcached.puts "stats cachedump #{slab_id} 100"
items << memcached.waitfor(/./).split("\n")[0].split(' ', 3)[1]
items
end
@gfa from the first couple of lines of the source in http://www.darkcoding.net/software/memcached-list-all-keys/
:
"In the general case, there is no way to list all the keys that a memcached instance is storing. You can, however, list something like the first 1Meg of keys, which is usually enough during development."
@bkimble thanks for this!
Here's yet another fork of your original gist:
https://gist.github.com/ginjo/aaa4a4395fcdbbad3d29
This revision adds a couple of features for working with cache item rows as ruby objects.
- Isolates the cache item query and enumeration into a method that yields to an optional block (passing all elements of cache item. You can do anything with the block - print rows, return rows as an array of hashes, return a single 'matching' row, etc..
- Accepts a list of servers and runs the above enumerator for all items found on every server.
I've included a method that mimics your original formatted printing, plus a method that returns all found cache items as an array of hashes.
Unfortunately it does not work when the server has several 100 thousand keys...
excellent, thanks a lot!
Nice 😮 . I have customized this script for dump and restore using dalli
gem.
https://gist.github.com/javanux/dbd005ef50ae4b8cea213af058abf771
I'm fairly certain this does not work on memcachier.
I successfully telneted in to our production servers and tried issues the stats cachedump command and got:
CLIENT_ERROR unsupported stats command
bingo! 👍
Thank you! This help me to solve problem of infinite caching(no expiration).
Hello, @bkimble. I have adjusted your Gist a bit, see: https://gist.github.com/archan937/f441146e1aeaa9da138e337514ee4000
The logic is in a method called keys
and you can pass your memcached host (or a list of hosts) and an optional pattern to filter on keys.
Also, prints the slab IDs with all of its keys.
Unfortunately it does not work when the server has several 100 thousand keys...
Same here. Does not work with 60k keys.
does it list all the keys or most of them?