When using the write cache and setting setAutoFlush() to false, is there a
risk of data loss, even if WAL is enabled ?

On Mon, Jun 20, 2011 at 12:27 PM, Jeff Whiting <je...@qualtrics.com> wrote:

> There is the possibility that your keys have the same timestamp --
> especially if you are running multi-threaded.  If the puts are buffered then
> it isn't outside the realm of possibility that they are executed within the
> same millisecond.  If you have the same keys for multiple puts you would
> "loose" data as you describe because it would just update the row rather
> than inserting a new one.
>
> ~Jeff
>
>
> On 6/20/2011 1:16 PM, Laurent Hatier wrote:
>
>> I think that there is a solution in your link, i will check it ! :)
>>
>> 2011/6/20 Laurent Hatier<laurent.hatier@gmail.**com<laurent.hat...@gmail.com>
>> >
>>
>>  my keys are the moment where the data is inserted into HBase (so
>>> System.currentTimeMillis()***1000). As you can see, i use the put method
>>> which
>>> insert data... there is an another way to insert data ?
>>>
>>>
>>> 2011/6/20 Doug 
>>> Meil<doug.meil@**explorysmedical.com<doug.m...@explorysmedical.com>
>>> >
>>>
>>>  Look here in the HBase book for these, and other, tips.
>>>>
>>>> http://hbase.apache.org/book.**html#performance<http://hbase.apache.org/book.html#performance>
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: jdcry...@gmail.com [mailto:jdcry...@gmail.com] On Behalf Of
>>>> Jean-Daniel Cryans
>>>> Sent: Monday, June 20, 2011 2:03 PM
>>>> To: user@hbase.apache.org
>>>> Subject: Re: Insert a lot of data in HBase
>>>>
>>>> 4M is small data :)
>>>>
>>>> Could there be an overlap in the keys? Are you disabling autoflush and
>>>> not
>>>> flushing the write buffer? These are common errors/misconceptions.
>>>>
>>>> J-D
>>>>
>>>> On Mon, Jun 20, 2011 at 10:36 AM, Laurent Hatier<
>>>> laurent.hat...@gmail.com>  wrote:
>>>>
>>>>> Hi all,
>>>>>
>>>>> I'm new in HBase. I want to insert 4'000'000 rows in HBase (each row
>>>>> has 4 columns). I have already looked the HBase wiki to insert data,
>>>>> but i've a problem : i loss data. When i do a COUNT with the shell,
>>>>> there is approximativly 1'500'000 in the DB...
>>>>> I've tested to create multiple Put and insert it with a List, i've
>>>>> already tested a simple Put with four add functions, open and close
>>>>> the socket it each time i put the line or i read the file don't run...
>>>>> If anyone have an idea.
>>>>>
>>>>> here we go my code if you want to see :
>>>>>
>>>>> List<Put>  arrayPut = new ArrayList<Put>();
>>>>>
>>>>> arrayPut.add(new Put(Bytes.toBytes(id)));
>>>>> arrayPut.get(arrayPut.size() - 1).add(FAMILY_GEOLOC, QUALIFIER_START,
>>>>> Bytes.toBytes(tStart)); arrayPut.add(new Put(Bytes.toBytes(id)));
>>>>> arrayPut.get(arrayPut.size() - 1).add(FAMILY_GEOLOC, QUALIFIER_END,
>>>>> Bytes.toBytes(tEnd)); arrayPut.add(new Put(Bytes.toBytes(id)));
>>>>> arrayPut.get(arrayPut.size() - 1).add(FAMILY_GEOLOC,
>>>>> QUALIFIER_COUNTRY, Bytes.toBytes(countryCode)); arrayPut.add(new
>>>>> Put(Bytes.toBytes(id)));
>>>>> arrayPut.get(arrayPut.size() - 1).add(FAMILY_GEOLOC, QUALIFIER_REGION,
>>>>> Bytes.toBytes(regionCode)); table.put(arrayPut);
>>>>>
>>>>> --
>>>>> Laurent HATIER
>>>>> Étudiant en 2e année du Cycle Ingénieur ą l'EISTI
>>>>>
>>>>>
>>>
>>> --
>>> Laurent HATIER
>>> Étudiant en 2e année du Cycle Ingénieur ą l'EISTI
>>>
>>>
>>
>>
> --
> Jeff Whiting
> Qualtrics Senior Software Engineer
> je...@qualtrics.com
>
>

Reply via email to