Hi Mateusz,

Are you using the default hypertable.cfg file or did you make custom
modifications.  This error will occur if the CellStoreScanner encounters a
CellStore block whose size is larger than the amount of memory dedicated to
the block cache.  Did you tweak either of the following two properties:

Hypertable.RangeServer.BlockCache.MaxMemory
Hypertable.RangeServer.CellStore.DefaultBlockSize

Or is it possible, that you have inserted an object that compressed is
larger than 200MB?  If this is a possibility, then do the following:

1. Modify the HT_FATALF statement on line 372 of CellStoreScannerV0.cc to be
this:

          HT_FATALF("Problem checking out block from cache file=%s
file_id=%d, "
                    "offset=%u", m_cell_store_ptr->get_filename().c_str(),
m_file_id, m_block.offset);

Then re-run your test and when it crashes again, it will print out the name
of the offending CellStore.  Then you can run the following command:

~/hypertable/0.9.2.2/bin/csdump <name-of-cellstore>

This will dump out the block index of the cell store.  See if there are any
blocks whose size is larger than your configured BlockCache (200MB by
default).

- Doug

On Fri, Mar 6, 2009 at 4:07 PM, Mateusz Berezecki <[email protected]>wrote:

>
> Hi,
>
> I'm again seeing this problem that I've experienced on 0.9.2.1:
>
> 1236384164 FATAL Hypertable.RangeServer :
>
> (/home/mateusz/hypertable/src/cc/Hypertable/RangeServer/CellStoreScannerV0.cc:372)
> Problem checking out block from cache file_id=29, offset=24492939
>
> What I am doing is a TFIDF over a text corpus using Hypertable as a
> backend. I'm doing cross table lookups, ie. looking up one value in
> table A and using it as a key for table B. I think I might be
> hammering the database too heavily, but who knows, maybe there in fact
> is some problem?
>
> My setup is: 4 cpu xeon, 4gb ram, 60gb sata disk with hypertable
> 0.9.2.2 with local broker. Did anyone encounter this behavior?
>
> Mateusz
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"Hypertable Development" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/hypertable-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to