Way to much import errors

[dendehun]

In the past I often got a lot of import errors when importing my mined hands.
This week I started with a new fresh database, assuming my database was corrupt.

However, today I have imported around 200k hands.
Around 50k of them resulted in import errors.

My import error log states a lot of these errors:
29SH2008.12.0569.txt ERROR: 25P02: current transaction is aborted, commands ignored until end of transaction block

Since it's not the database (anymore), and not the hh's themself, what can cause these problems, and more important, how can I solve this?

I use 1.07.01, just tried to import yesterday's hands resulting in only import errors, no hands actually imported (out of 130K hands)!!

Please advice

[ilovewta]

:eek:

[Al1]

Can you reinstall postgreSQL and tell me if it changes something (create a new DB after reinstalling postgreSQL). http://208.109.95.123/faq/afmviewfaq.aspx?faqid=72

[dendehun]

After a lot of hours re-importing hands in my new postgresql install & new hem database, new imports work without import errors now.

However, this was the same when I created a new database earlier this week.
I'm not planning to create a new database every week, so I'm still very eager to know what could cause these problems in the first place.

In the meantime I will check my logs frequently and let you know when it fails again (somehow I have no confidence that this solved my problems permanently).

[morny]

The 2 most common causes of this problems are 1) powering off your computer or having your computer lose power while the database is doing something and 2) an old drive with bad sectors on it. Basically what happens is a small portion of a file gets slightly corrupt and then when you reboot and windows goes into that checkdsk procedure it finds the bad section and, thinking it is doing a good thing, removes it. Postgres then loads the table and a portion of it is gone so it doesn’t like that and blocks access to the entire table. Normally when this happens it is on a completely useless file that you'll never even notice is now gone and quite often it will happen to an index in the DB which can easily be rebuilt but in your case it happened to one or more of the tables.

So, to protect against this

1) keep you hands histories in case you do need to reimport
2) use a power bar and avoid hard shutdowns (when you flick the power switch on the pc)
3) if your drive is old consider replacing it with a new one. They are cheap and much faster now than even a few years ago.


See Roys quote on corrupt databases