clustering for 32 hours

PostgreSQL is the database server used to store information. Do you have a question or are you having problem with PostgreSQL? If so, post them here.

Moderator: Moderators

clustering for 32 hours

Postby onaleeps » Wed Feb 11, 2009 1:05 am

I just manually uploaded 2 million hands. It took about 11 hours to upload them, but they finally finished. That was yesterday. It has now been clustering the database for 32 hours 21 min. I've tried to be patient, waiting for it to finish, but is this possible? I don't want to cancel the clustering, but I'm worried there may be a problem. Is it possible for it to take this long?

Thank you.
onaleeps
 
Posts: 16
Joined: Sun Jan 25, 2009 2:54 am

Re: clustering for 32 hours

Postby kraada » Wed Feb 11, 2009 10:42 am

How much free space do you have on the hard drive in question? If you don't have twice the size of the database available as free space clustering can take significantly longer. For a 2M hand database, you'll probably want something like 40G of hard drive space free on that hard drive, assuming you don't have any other data in that database. If you only have something like 10G free, clustering could take a very long time to finish (as you've seen).
kraada
Moderator
 
Posts: 54430
Joined: Wed Mar 05, 2008 2:32 am
Location: NY

Re: clustering for 32 hours

Postby onaleeps » Wed Feb 11, 2009 12:12 pm

Thank you for your reply. You are right, the hard drive space was low. Nonetheless, during the night last night, it finally finished. Now, the database queries and reports are excruciatingly slow. So, I would like to split the database up. I've read about making two databases, one for hands that I'm involved in, and another for downloaded hands that I'm not involved in. If I already have one database with all the hands in it, is there a way to filter the database for only hands I was involved in and then split the database into two?

Thank you.
onaleeps
 
Posts: 16
Joined: Sun Jan 25, 2009 2:54 am

Re: clustering for 32 hours

Postby kraada » Wed Feb 11, 2009 1:29 pm

Not in one step, but it is possible.

First what you'll want to do is export all hands from the database (Database --> Export) and save them to a new directory. Then you'll want to click Datbase --> Purge... and purge all observed hands from the database. Go to the Advanced tab, make sure your screen name is selected properly, and check "Purge Observed Hands" and click OK. Depending on how large your database is, this can take quite a while (you might want to let it run overnight). Once that finishes, perform housekeeping on your database. This should go much faster as your database will now be a lot smaller. Once that completes, create a new database (from the Database Management window) and set it to active (but not default, you'll probably want your played hands database as default), then go to the Manual Import tab, and import all of your hand histories into that database. Once they're imported, go back to the Purge window and purge all hands you've played (click on yourself then highlight all of your sessions and click "purge sessions"). This may also take a while, but not as long as the first purge. Once this purge completes, again you should run Housekeeping this time just on the new database.

The last thing you should do is go to the Auto Import tab, go to the Preferences tab on the right and make sure played hands import into your played database and observed hands go to your observed database; then in the future they'll be sorted automatically when auto importing. (If you manually import large chunks of datamined data, you'll want to set your datamined database to active as Manual Import always imports to the active database.)
kraada
Moderator
 
Posts: 54430
Joined: Wed Mar 05, 2008 2:32 am
Location: NY

Re: clustering for 32 hours

Postby onaleeps » Wed Feb 11, 2009 5:27 pm

Thanks again. O.k., continuing (sorry, I think this will be my last one...)

I've read your instructions on splitting the database and think I can follow them. But first I want to move my whole database over to another computer, one with more resources (faster hard drive RPM, etc). What is the best way to do that? I found instructions on the Poker Tracker website, but they don't make any sense. The beginning of the instructions is copied below. It says "Open the Utilities/PostgreSQL Database Funcitons window". Where the heck is that?

Thank you.


**** Below are the instructions I found on the Web, but they didn't make sense ******

To do the backup (on the old computer):

1. Open the Utilities/PostgreSQL Database Functions window.
2. Click the "Backup" button to create a compressed backup file of the selected PostgreSQL database. If you have more than one database, then you must choose each PostgreSQL database from the dropdown list at the top left corner of the window and perform the backup for EACH database. The backup being done is only for the selected database and NOT for all your databases. So if you have created 3 PostgreSQL databases, you need to perform the backup function 3 times, once for each database.
onaleeps
 
Posts: 16
Joined: Sun Jan 25, 2009 2:54 am

Re: clustering for 32 hours

Postby kraada » Wed Feb 11, 2009 5:36 pm

Those are instructions for PokerTracker 2, not PokerTracker 3.

To backup your database in PT3, please see the instructions in this post.
kraada
Moderator
 
Posts: 54430
Joined: Wed Mar 05, 2008 2:32 am
Location: NY


Return to PostgreSQL [Read Only]

Who is online

Users browsing this forum: No registered users and 6 guests