Skip to content

Conversation

BohuTANG
Copy link
Contributor

[summary]
When we do hot backup like below:

  1. take checkpoint lock
  2. lock table for read
  3. wirte binary log postion
  4. copy redo log to backup dir
  5. unlock tables
  6. start to copy datas files
    6.1) write data by another client
    6.2) copy file1 to backup dir
  7. release checkpoint lock
    Between 6.1) and 6.2) the FT may does ft_close and update FT header to new, the new writes are visible in the backup instance, this broke the consistency.
    It would be better for TokuDB to update FT header only in checkpoint, keep all FT files consistency.

Copyright (c) 2015, BohuTANG
All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

  1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
  2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

[summary]
When we do hot backup like below:
1) take checkpoint lock
2) lock table for read
3) wirte binary log postion
4) copy redo log to backup dir
5) unlock tables
6) start to copy datas  files
   6.1) write data by another client
   6.2) copy file1 to backup dir
7) release checkpoint lock
Between 6.1) and 6.2) the FT may does ft_close and update FT header to new, the new writes are visible in the backup instance, this broke the consistency.
It would be better for TokuDB to update FT header only in checkpoint, keep all FT files consistency.

Copyright (c) 2015, BohuTANG
All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
@george-lorch
Copy link

This is a very nice idea. Thank you for the contribution. I need a little time to analyze and review the idea before merging.

@kuszmaul
Copy link
Contributor

Clever. Assuming that works, maybe we can avoid cloning nodes during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" [email protected]
wrote:

This is a very nice idea. Thank you for the contribution. I need a little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
#323 (comment).

@kuszmaul
Copy link
Contributor

I guess I don't see how to use this idea to reduce checkpoint variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't implemented it) is
that the Toku hot backup can, in principle do incremental backup. Is that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" [email protected] wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" [email protected]
wrote:

This is a very nice idea. Thank you for the contribution. I need a little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
#323 (comment).

@BohuTANG
Copy link
Contributor Author

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we can
get the binlog info and build a slave instance, then do replication with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]> wrote:

I guess I don't see how to use this idea to reduce checkpoint variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't implemented it) is
that the Toku hot backup can, in principle do incremental backup. Is that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');>
wrote:

This is a very nice idea. Thank you for the contribution. I need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
#323 (comment).


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh

@kuszmaul
Copy link
Contributor

I'd like to better understand what toku hot backup cannot do. Doesn't it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" [email protected] wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we can
get the binlog info and build a slave instance, then do replication with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't implemented it)
is
that the Toku hot backup can, in principle do incremental backup. Is that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');>
wrote:

This is a very nice idea. Thank you for the contribution. I need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<#323 (comment)
.


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).

@BohuTANG
Copy link
Contributor Author

If the master instance is very busy, how TokuDB hotbackup create
replication slaves?
From the codes, it is only for datas backup can not create slave easily.
But its idea is very cool!

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]> wrote:

I'd like to better understand what toku hot backup cannot do. Doesn't it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we can
get the binlog info and build a slave instance, then do replication with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional
backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't implemented
it)
is
that the Toku hot backup can, in principle do incremental backup. Is
that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>>
wrote:

This is a very nice idea. Thank you for the contribution. I need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh

@prohaska7
Copy link
Contributor

Hotbackup can backup multiple directories. If one backs up the data dirs
and the binlog dirs, then one can compute the master binlog position from
the backed up binlog that matched the data files.
On Sep 16, 2015 8:38 AM, "BohuTANG" [email protected] wrote:

If the master instance is very busy, how TokuDB hotbackup create
replication slaves?
From the codes, it is only for datas backup can not create slave easily.
But its idea is very cool!

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]> wrote:

I'd like to better understand what toku hot backup cannot do. Doesn't it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we
can
get the binlog info and build a slave instance, then do replication
with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional
backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't implemented
it)
is
that the Toku hot backup can, in principle do incremental backup. Is
that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes
during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>>
wrote:

This is a very nice idea. Thank you for the contribution. I need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
<#323 (comment)
.


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).

@BohuTANG
Copy link
Contributor Author

Rich,
It's hard to get the point-in-time for replication with hotbackup, during
copying all files is active.

On Wednesday, September 16, 2015, Rik Prohaska [email protected]
wrote:

Hotbackup can backup multiple directories. If one backs up the data dirs
and the binlog dirs, then one can compute the master binlog position from
the backed up binlog that matched the data files.
On Sep 16, 2015 8:38 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

If the master instance is very busy, how TokuDB hotbackup create
replication slaves?
From the codes, it is only for datas backup can not create slave easily.
But its idea is very cool!

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

I'd like to better understand what toku hot backup cannot do. Doesn't
it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we
can
get the binlog info and build a slave instance, then do replication
with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional
backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't
implemented
it)
is
that the Toku hot backup can, in principle do incremental backup.
Is
that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes
during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>>
wrote:

This is a very nice idea. Thank you for the contribution. I
need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh

@prohaska7
Copy link
Contributor

Why? When Tokutek hot backup finishes, one has a consistent snapshot of the
tokudb data files, tokudb log files, and the binlog (as long as
sync_binlog=1). One can compute master position from the binlog.
Rich,
It's hard to get the point-in-time for replication with hotbackup, during
copying all files is active.

On Wednesday, September 16, 2015, Rik Prohaska [email protected]
wrote:

Hotbackup can backup multiple directories. If one backs up the data dirs
and the binlog dirs, then one can compute the master binlog position from
the backed up binlog that matched the data files.
On Sep 16, 2015 8:38 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

If the master instance is very busy, how TokuDB hotbackup create
replication slaves?
From the codes, it is only for datas backup can not create slave easily.
But its idea is very cool!

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:

I'd like to better understand what toku hot backup cannot do. Doesn't
it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that we
can
get the binlog info and build a slave instance, then do replication
with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional
backup
work. I'd be interested in a performance and usability comparison.

One difference that does show up today (since we haven't
implemented
it)
is
that the Toku hot backup can, in principle do incremental backup.
Is
that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes
during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>>
wrote:

This is a very nice idea. Thank you for the contribution. I
need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).

@BohuTANG
Copy link
Contributor Author

Looks good to me, I will have a try.

On Thursday, September 17, 2015, Rik Prohaska [email protected]
wrote:

Why? When Tokutek hot backup finishes, one has a consistent snapshot of the
tokudb data files, tokudb log files, and the binlog (as long as
sync_binlog=1). One can compute master position from the binlog.
Rich,
It's hard to get the point-in-time for replication with hotbackup, during
copying all files is active.

On Wednesday, September 16, 2015, Rik Prohaska <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');>
wrote:

Hotbackup can backup multiple directories. If one backs up the data dirs
and the binlog dirs, then one can compute the master binlog position from
the backed up binlog that matched the data files.
On Sep 16, 2015 8:38 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

If the master instance is very busy, how TokuDB hotbackup create
replication slaves?
From the codes, it is only for datas backup can not create slave
easily.
But its idea is very cool!

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:

I'd like to better understand what toku hot backup cannot do. Doesn't
it
back up the binlog too?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 8:07 AM, "BohuTANG" <[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>> wrote:

Hi,
We have hack TokuDB hot backup to Xtrabackup as above said.
It's all about Linux 'cp' command and the important thing is that
we
can
get the binlog info and build a slave instance, then do replication
with
master, TokuDB hotbackup can't do that.

On Wednesday, September 16, 2015, Bradley C. Kuszmaul <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>> wrote:

I guess I don't see how to use this idea to reduce checkpoint
variability.
Too early in the morning I guess.

It is still a clever idea that seems like it can make traditional
backup
work. I'd be interested in a performance and usability
comparison.

One difference that does show up today (since we haven't
implemented
it)
is
that the Toku hot backup can, in principle do incremental backup.
Is
that
important?

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 5:54 AM, "Bradley C. Kuszmaul" <
[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>');>> wrote:

Clever. Assuming that works, maybe we can avoid cloning nodes
during
checkpoint, and incur checkpoint overhead only for the headers.

Bradley C Kuszmaul - via snartphone
On Sep 16, 2015 12:07 AM, "georgelorchpercona" <
[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');
<javascript:_e(%7B%7D,'cvml','[email protected]
javascript:_e(%7B%7D,'cvml','[email protected]');');>');>');>>
wrote:

This is a very nice idea. Thank you for the contribution. I
need a
little
time to analyze and review the idea before merging.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
<
#323 (comment)
.


Reply to this email directly or view it on GitHub
<
#323 (comment)
.

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
<#323 (comment)
.


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh


Reply to this email directly or view it on GitHub
#323 (comment).


Reply to this email directly or view it on GitHub
#323 (comment).

BohuTANG

"The great artist is the simplifier."--- Vincent Van Gogh

@BohuTANG
Copy link
Contributor Author

This patch has some bugs when table create and close less than a checkpoint interval, will fixed.

@BohuTANG BohuTANG closed this Sep 21, 2015
@BohuTANG BohuTANG deleted the feature_keep_ft_files_consistency_FT701 branch September 29, 2015 09:18
@BohuTANG
Copy link
Contributor Author

BohuTANG commented Oct 3, 2015

Moved to #331

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants