Erlang unique_integer returns the same integer after restart

Question

I need to number my records so that I can always compare what came "later" from what came "before".

For this I used the unique_integer function:

Tag = erlang:unique_integer([monotonic]),

and then I would compare any two events:

Event1.tag < Event2.tag

would mean that Event1 came before Event2.

However, when I restart the erlang node, it starts numbering from the beginning again:

event1  -5555
event2  -5554
event3  -5553

-- restart the system

event4  -5555
event5  -5554
event6  -5553

Now my sorting is not working as intended.

Is there any way to generate long sequential number that picks up where it left off after restarts?

P.S.: If it is useful, I am using mnesia to store my events, and so I need an increasing row number for record.

Answer 1

This is the feature I was looking for: mnesia:dirty_update_counter.

Thanks to this excellent post on the matter:

mnesia:dirty_update_counter(unique_ids, record_type, 1), 
io:format("Id => ~p~n", [Id]), 

Id1 = mnesia:dirty_update_counter(unique_ids,record_type, 1), 
io:format( "Id => ~p~n", [Id1]), 

Id2 = mnesia:dirty_update_counter(unique_ids, another_type, 1), 
io:format("Id => ~p~n", [Id2]),

The output you will get is

Id => 1 
Id => 2 
Id => 1

A single table can be used to generate the unique ids for other tables. In this example, unique ids are generated for record_type and another_type. Posted by Dude From Mangalore at 3:19 PM

Answer 2

From the documentation:

erlang:unique_integer() -> integer()

Generates and returns an integer unique on current runtime system instance. The same as calling erlang:unique_integer([]).

So what you're seeing is consistent with the defined behavior.

---

I'm not clear what you need this for, and the correct solution depends on your exact needs. Assuming you have events coming in rapidly and you're planning to use ordered_set to store them in Mnesia by arrival time I can think of a few approaches:

• Have a single process in charge of inserts. It can initialize with the max ID from the table and increment on each insert.
• Generate a key with erlang:system_time(nano_seconds) that's unique-ish
• Generate a key with {erlang:system_time(nano_seconds), make_ref()} to ensure uniqueness even when the timestamps are the same