libuuid generates one of several types of uuid.

One simply makes all bits random from /dev/urandom. Collisions are unlikely but still possible.

The time one allocates some bits to the mac and others to the unix timestamp, then the rest is sequential. That one means different servers will never generate collisions (assuming all use the same scheme), and collisions can't happen within the program on the server. That can be made between multiple programs using the library too.

http://linux.die.net/man/3/uuid_generate

Personally I would prefer the later but for privacy reasons (the mac) the first is the libraries default. Not sure which FS uses...




On Sunday, May 19, 2013, Jeff Leung wrote:

It depends on what implementation of the UUID library the build system has detected.

If you’ve had the libuuid from e2fsprogs (Linux at the very least) installed and headers present, the build system will link against that. That implementation generates random UUID’s for the most part. If you didn’t have libuuid and its headers installed, the build system will use APR’s own implementation for generating the UUID’s; those generated UUID’s are known to be sequential in nature.

 

From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Telecube - John
Sent: Sunday, May 19, 2013 6:58 AM
To: freeswitch-users@lists.freeswitch.org
Subject: [Freeswitch-users] UUID - how unique is it?

 

How unique is uuid in the cdr records?

Coming from asterisk where uid is about as unique as a black bowling ball....

Can I count on uuid to be totally unique  in my billing records?

--
Regards,
John Matich
Telecube Pty Ltd
www.telecube.com.au
Ph: 13CUBE (132823)