The Milkfish "Availability-and-Utilisation-in-a-Distributed-Infrastructure Tool"

or short: A.U.D.I.T.

This system was a by-product of the Milkfish developers system stability evaluations...and now it is available to every Milkfish/Openwrt/DD-WRT/Linux user.

Procedure:

1. Registration - The user registers as a person at forum.milkfish.org
2. Installation - The user installs the milkfish-services package and reboots the Openwrt router (not necessary for DD-WRT routers)
3. Enabling - The user enables the audit process on the router (i.e. milkfish_audit=on in NVRAM)
4. Activation - The user activates the router for auditing (i.e. submitting a router specific ID)
5. Updating - The router periodically updates the server at milkfish.org with its current performance
6. Auditing - The user is auditing the routers performance by reviewing the cumulated results

If one of the steps is not performed or if a registration or activation is withdrawn by the user, the server at milkfish.org can not perform any auditing. The same holds for the case when auditing is disabled on the router itself (i.e. milkfish_audit=off in NVRAM).

Here's an example for a manual Audit Activation on a router:

 root@openwrt:~# milkfish_services status
 The Milkfish Router Services
 Checking router status at milkfish.org (https)...
 Milkfish Services - Router Status
 User at forum.milkfish.org: REGISTERED
 Milkfish Audit: INACTIVE
 Current Router ID: INVALID

 root@openwrt:~# milkfish_services audit activate
 The Milkfish Router Services
 Activating the Milkfish Audit service (https)...
 Activation completed

 root@openwrt:~# milkfish_services status
 The Milkfish Router Services
 Checking router status at milkfish.org (https)...
 Milkfish Services - Router Status
 User at forum.milkfish.org: REGISTERED
 Milkfish Audit: ACTIVE
 Current Router ID: VALID

 root@openwrt:~#

Audit Results are calculated and prepared for download by the Milkfish audit server. Please note, that after activating audit it may take up to an hour until first results can be reviewed. Links to the respective graphs are available through the Milkfish web interface on Openwrt routers (for DD-WRT routers see descriptions below).

Here is an example for a text-based Audit Results check:

 root@openwrt:~# milkfish_services audit results
 The Milkfish Router Services
 Requesting Milkfish Audit results (http)...
 Milkfish Router Audit Results
 Registered auditor : fronce 
 Audit period       : 14.19 d
 Router availability: 100%
 SER availability   : 99.15%
 Router utilisation : 0.196%
 Router status      : online
 Wed Sep 13 23:53:09 CEST 2006
 root@openwrt:~#

Audit Results are also available within a router specific signet like this or the one further below:



If you have an active Audit account you can get your Router ID (RID) from the Audit Results Overview and add "-liner.png", "-signet.png" or "-uptime.png" to it. This way you can integrate the graphic into your own HTML Code on your website or your forum, if you like. The location of all results and graphics is: http://www.milkfish.org/audit/results/ (your Router ID) "-" "result"."filetype"

Example:



To get the Link to your own result graphic, first click on one of the the graphic examples above and copy the link location and then change the RID-part of the file name to your own RID. You can find your own RID in the back of the table here: http://www.milkfish.org/audit/results/

It is then possible to link the signet above to a corresponding and more detailed graphical uptime plot: (see on top of page)

How it works:
The router under audit sends, say, every half an hour a message to the audit server saying "I'm fine. I was so since my last message."
Over a long period of time the server can thereby determine at what times the router was fine, meaning working reliably, and at what times it was not. By comparing these two amounts of time, the routers availability is calculated.
Example: If the router was audited for 20 days, being online for 10 days and offline for the other 10 days, then the router was 50% available during these 20 days. It is obvious that this availability value itself becomes more reliable the longer we look at the router.

Reasons for the Audit Service

Goal: Making the reliability of complex open source driven systems easily reviewable.

Why? Because we want to know how reliable them hobbyists systems are at the end of the day.

What? Why, again? Because...

1. when communication systems need to be reliable then toys need to be so even more
2. system requirements are similar and shared among online technologies
3. you can buy books about it