hsync in hadoop 2.0?

[lars hofhansl <lh...@yahoo.com>]

HDFS-744 adds support for true durable sync for DFSOutputStream.hsync().
This is useful for HBase (among others) and I created a matching patch in HBASE-5954.

I would like to port this back into Hadoop 2.0 and maybe even 1.0.
A 2.0 patch is already attached to HDFS-744. (The 1.0 patch would be quite different, but the underlying implementation would be the same)


If I understand correctly Hadoop 2.0 is not out of alpha, yet.

Thoughts?

-- Lars

Re: hsync in hadoop 2.0?

[Eli Collins <el...@cloudera.com>]

+1

IMO we should hold off on branch-1 unless people really need it.

On Mon, Jun 4, 2012 at 10:33 AM, Tsz Wo Sze <sz...@yahoo.com> wrote:
> +1
>
> Since no one is objecting, I will merge HDFS-744 to 2.0.
>
> For 1.x, it needs more thoughts because it is likely to be an incompatible change.
>
>
> Tsz-Wo
>
>
>
>
> ________________________________
>  From: Todd Lipcon <to...@cloudera.com>
> To: hdfs-dev@hadoop.apache.org
> Cc: lars hofhansl <lh...@yahoo.com>
> Sent: Friday, June 1, 2012 10:04 AM
> Subject: Re: hsync in hadoop 2.0?
>
> +1, it's protocol-compatible, I don't see any reason not to put it in 2.0.x.
>
> On Fri, Jun 1, 2012 at 5:02 AM, Andrew Purtell <ap...@apache.org> wrote:
>> +1 for 2.0
>>
>> For those trying to stabilize that for deployment early next year, commit
>> of this only to trunk won't help as much, though a private backport could
>> happen in that case.
>>
>>    - Andy
>>
>> On Friday, June 1, 2012, lars hofhansl wrote:
>>
>>> HDFS-744 adds support for true durable sync for DFSOutputStream.hsync().
>>> This is useful for HBase (among others) and I created a matching patch in
>>> HBASE-5954.
>>>
>>> I would like to port this back into Hadoop 2.0 and maybe even 1.0.
>>> A 2.0 patch is already attached to HDFS-744. (The 1.0 patch would be quite
>>> different, but the underlying implementation would be the same)
>>>
>>>
>>> If I understand correctly Hadoop 2.0 is not out of alpha, yet.
>>>
>>> Thoughts?
>>>
>>> -- Lars
>>>
>>
>>
>> --
>> Best regards,
>>
>>   - Andy
>>
>> Problems worthy of attack prove their worth by hitting back. - Piet Hein
>> (via Tom White)
>
>
>
> --
> Todd Lipcon
> Software Engineer, Cloudera

Re: hsync in hadoop 2.0?

[Tsz Wo Sze <sz...@yahoo.com>]

+1

Since no one is objecting, I will merge HDFS-744 to 2.0.

For 1.x, it needs more thoughts because it is likely to be an incompatible change.


Tsz-Wo




________________________________
 From: Todd Lipcon <to...@cloudera.com>
To: hdfs-dev@hadoop.apache.org 
Cc: lars hofhansl <lh...@yahoo.com> 
Sent: Friday, June 1, 2012 10:04 AM
Subject: Re: hsync in hadoop 2.0?
 
+1, it's protocol-compatible, I don't see any reason not to put it in 2.0.x.

On Fri, Jun 1, 2012 at 5:02 AM, Andrew Purtell <ap...@apache.org> wrote:
> +1 for 2.0
>
> For those trying to stabilize that for deployment early next year, commit
> of this only to trunk won't help as much, though a private backport could
> happen in that case.
>
>    - Andy
>
> On Friday, June 1, 2012, lars hofhansl wrote:
>
>> HDFS-744 adds support for true durable sync for DFSOutputStream.hsync().
>> This is useful for HBase (among others) and I created a matching patch in
>> HBASE-5954.
>>
>> I would like to port this back into Hadoop 2.0 and maybe even 1.0.
>> A 2.0 patch is already attached to HDFS-744. (The 1.0 patch would be quite
>> different, but the underlying implementation would be the same)
>>
>>
>> If I understand correctly Hadoop 2.0 is not out of alpha, yet.
>>
>> Thoughts?
>>
>> -- Lars
>>
>
>
> --
> Best regards,
>
>   - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)



-- 
Todd Lipcon
Software Engineer, Cloudera

Re: hsync in hadoop 2.0?

[Todd Lipcon <to...@cloudera.com>]

+1, it's protocol-compatible, I don't see any reason not to put it in 2.0.x.

On Fri, Jun 1, 2012 at 5:02 AM, Andrew Purtell <ap...@apache.org> wrote:
> +1 for 2.0
>
> For those trying to stabilize that for deployment early next year, commit
> of this only to trunk won't help as much, though a private backport could
> happen in that case.
>
>    - Andy
>
> On Friday, June 1, 2012, lars hofhansl wrote:
>
>> HDFS-744 adds support for true durable sync for DFSOutputStream.hsync().
>> This is useful for HBase (among others) and I created a matching patch in
>> HBASE-5954.
>>
>> I would like to port this back into Hadoop 2.0 and maybe even 1.0.
>> A 2.0 patch is already attached to HDFS-744. (The 1.0 patch would be quite
>> different, but the underlying implementation would be the same)
>>
>>
>> If I understand correctly Hadoop 2.0 is not out of alpha, yet.
>>
>> Thoughts?
>>
>> -- Lars
>>
>
>
> --
> Best regards,
>
>   - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)



-- 
Todd Lipcon
Software Engineer, Cloudera

Re: hsync in hadoop 2.0?

[Andrew Purtell <ap...@apache.org>]

+1 for 2.0

For those trying to stabilize that for deployment early next year, commit
of this only to trunk won't help as much, though a private backport could
happen in that case.

    - Andy

On Friday, June 1, 2012, lars hofhansl wrote:

> HDFS-744 adds support for true durable sync for DFSOutputStream.hsync().
> This is useful for HBase (among others) and I created a matching patch in
> HBASE-5954.
>
> I would like to port this back into Hadoop 2.0 and maybe even 1.0.
> A 2.0 patch is already attached to HDFS-744. (The 1.0 patch would be quite
> different, but the underlying implementation would be the same)
>
>
> If I understand correctly Hadoop 2.0 is not out of alpha, yet.
>
> Thoughts?
>
> -- Lars
>


-- 
Best regards,

   - Andy

Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)