cluster stops when not enough hard space

[PT23]

Hi I have a database where I just importet like 5mio hands, thats aprox. 40Gb in the SQL Database.
Now, postgreSQL is intalled on a partition where there are only 7Gb left and when I now make a cluster
postgreSQL makes temporary files that get rather large, so large that my hard space is not enough.
But instead of telling me this problem PT3 just stops the cluster and tells me it is done as if everything whent all right which cant be,
it took much too short time and happens exactly when the temps fill the left hard space and can't grow any bigger.
The outcome is that spadeeye cant read the database properly, it gets no stats.

What I'm probably gonna do is, wait for the next beta that enables me to purge selective data since I have a lot
of hands of the wrong limit in there, and then try again.
But perhaps you guys should address this and implement a kinda error message. And why does the cluster needs more
than 7 GB temporal space to cluster a 40GB database?

P.S.: What about disc deframentation? When I do a defragmentation of that partition after a cluster does that f*ck up the cluster or are theese two different shoes?

[WhiteRider]

I can't find the post now but I'm sure that Josh said yesterday that you should cluster and vacuum, then defrag, then analyse and build the cache.

I don't know enough about postgres to answer the rest of your points.

[PT23]

The post u are mentioning is the new beta 16 thread.
And there is no mention of the selective purge feature that was promissed for this release
or perhaps I got it wrong and it was ment for beta 17.
Well I still have the problem with the hard space so I guess I have to export all hands
deinstall everything and install and import on a larger partition.
Pritty anoying but I guess I have no other option.

[WhiteRider]

Selective Purge is listed as an upcoming feature, but not specifically for beta 16.
I expect it to be added within the next couple of betas - there may be another bug fix beta before the next few features, but keep an eye on the Development Timeline for Josh's updates.

[APerfect10]

If you know how to execute SQL commands via Pgadmin, you can individually cluster each table one by one which should limit the amount of temporary disk space required.

CLUSTER <table name>;

Repeat for every table.

Best regards,

Derek

PS. Selective purging was never scheduled for Beta 16. It will make it into a release by the end of September at the very latest...