postgres-xc-developers Mailing List for Postgres-XC

On Fri, Feb 10, 2012 at 1:46 PM, Michael Paquier
<mic...@gm...> wrote:
>
>
> On Fri, Feb 10, 2012 at 4:34 PM, Ashutosh Bapat
> <ash...@en...> wrote:
>>
>> Can initgtm be initdb with node type gtm or gtm-proxy. I know initdb and
>> initgtm functionality is almost orthogonal, but this approach will save code
>> copying in following respect.
>
> This will increase influence of XC code in initdb more than necessary, GTM
> and XC nodes are slightly different functionalities. By putting too much our
> hands in initdb code, I fear that future merges will become a pain and I
> believe we should avoid conflicts as much as possible.
>

Yeah, I think initdb and initgtm should remain as different binaries.
While I haven't looked at initgtm patch yet, I would imagine initdb
would do a lot more work than initgtm.

One of the things that we discussed in Dec is to pass a cluster UUID
to initgtm or let is generate one and use that to tie all other
pieces. This will ensure that we don't accidentally connect to a wrong
GTM or data node.

Thanks,
Pavan

-- 
Pavan Deolasee
EnterpriseDB http://www.enterprisedb.com

On Fri, Feb 10, 2012 at 4:34 PM, Ashutosh Bapat <
ash...@en...> wrote:

> Can initgtm be initdb with node type gtm or gtm-proxy. I know initdb and
> initgtm functionality is almost orthogonal, but this approach will save
> code copying in following respect.
>
This will increase influence of XC code in initdb more than necessary, GTM
and XC nodes are slightly different functionalities. By putting too much
our hands in initdb code, I fear that future merges will become a pain and
I believe we should avoid conflicts as much as possible.


>
> 1. there is a lot of process handling code in initgtm code, which might be
> common to initdb
>
pqsignal for instance, then only static functions.

> 2. we might use the same directory creation logic as initdb

Isn't it already the case?

>
>
> On Fri, Feb 10, 2012 at 12:54 PM, Koichi Suzuki <koi...@gm...>wrote:
>
>> Hi,
>>
>> Gtm.conf and gtm_proxy.conf file cannot be used as is.   This is
>> different from vanilla PG's initdb.   In vanilla PG, we can issue
>> "pg_ctl start" just after initdb, if people are happy with default
>> settings.
>>
>> In this implementation, we cannot issue "gtm_ctl start" right after
>> gtminit because users have to resolve mandatory settings such as
>> "host", "port", "gym_host" and "gtm_port" (the latter two are just for
>> gtm_proxy.conf).
>>
>> So it seems to me that gtminit just creates work director and copies
>> sample gtm.conf or gtm_proxy.conf and it seems to me that gtminit
>> doesn't help much.   Are there any other things which gtminit helps?
>>
>> Regards;
>> ----------
>> Koichi Suzuki
>>
>>
>>
>> 2012/2/10 Michael Paquier <mic...@gm...>:
>> > Hi all,
>> >
>> > Please find a patch to support a new module called initgtm.
>> > It is based on initdb, and initializes the data folder of GTM,
>> particularly
>> > useful to avoid copying configuration files directly from share folder
>> for
>> > GTM and GTM-Proxy, in case user has no access to the install repository.
>> > The spec of the module is as follows:
>> > $ initgtm --help
>> > initgtm initializes a GTM for Postgres-XC database cluster.
>> >
>> > Usage:
>> >   initgtm [NODE-TYPE] [OPTION]... [DATADIR]
>> >
>> > Options:
>> >  [-D, --pgdata=]DATADIR     location for this GTM node
>> >  [-Z]NODE-TYPE              can be "gtm" or "gtm_proxy"
>> > Less commonly used options:
>> >   -d, --debug               generate lots of debugging output
>> >   -n, --noclean             do not clean up after errors
>> >   -s, --show                show internal settings
>> >
>> > Other options:
>> >   -?, --help                show this help, then exit
>> >   -V, --version             output version information, then exit
>> >
>> > Comments are welcome.
>> > --
>> > Michael Paquier
>> > http://michael.otacoo.com
>> >
>> >
>> ------------------------------------------------------------------------------
>> > Virtualization & Cloud Management Using Capacity Planning
>> > Cloud computing makes use of virtualization - but cloud computing
>> > also focuses on allowing computing to be delivered as a service.
>> > http://www.accelacomm.com/jaw/sfnl/114/51521223/
>> > _______________________________________________
>> > Postgres-xc-developers mailing list
>> > Pos...@li...
>> > https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>> >
>>
>>
>> ------------------------------------------------------------------------------
>> Virtualization & Cloud Management Using Capacity Planning
>> Cloud computing makes use of virtualization - but cloud computing
>> also focuses on allowing computing to be delivered as a service.
>> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>> _______________________________________________
>> Postgres-xc-developers mailing list
>> Pos...@li...
>> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>>
>
>
>
> --
> Best Wishes,
> Ashutosh Bapat
> EntepriseDB Corporation
> The Enterprise Postgres Company
>
>


-- 
Michael Paquier
http://michael.otacoo.com

On Fri, Feb 10, 2012 at 4:24 PM, Koichi Suzuki <koi...@gm...> wrote:

> Hi,
>
> Gtm.conf and gtm_proxy.conf file cannot be used as is.   This is
> different from vanilla PG's initdb.   In vanilla PG, we can issue
> "pg_ctl start" just after initdb, if people are happy with default
> settings.
>
Same as current initdb with XC nodes.


>
> In this implementation, we cannot issue "gtm_ctl start" right after
> gtminit because users have to resolve mandatory settings such as
> "host", "port", "gym_host" and "gtm_port" (the latter two are just for
> gtm_proxy.conf).

There are still no default values for gtm_port and gtm_host as I know.
Using a configuration file without setting them now crashes GTM.

>


> So it seems to me that gtminit just creates work director and copies
> sample gtm.conf or gtm_proxy.conf and it seems to me that gtminit
> doesn't help much.   Are there any other things which gtminit helps?
>
1) Cluster bootstrap by using it to initialize GTM with a cluster ID.
2) Install a GTM folder for users who do not have an access

Still in all the cases a basic infrastructure is necessary.
-- 
Michael Paquier
http://michael.otacoo.com

Can initgtm be initdb with node type gtm or gtm-proxy. I know initdb and
initgtm functionality is almost orthogonal, but this approach will save
code copying in following respect.

1. there is a lot of process handling code in initgtm code, which might be
common to initdb
2. we might use the same directory creation logic as initdb

On Fri, Feb 10, 2012 at 12:54 PM, Koichi Suzuki <koi...@gm...>wrote:

> Hi,
>
> Gtm.conf and gtm_proxy.conf file cannot be used as is.   This is
> different from vanilla PG's initdb.   In vanilla PG, we can issue
> "pg_ctl start" just after initdb, if people are happy with default
> settings.
>
> In this implementation, we cannot issue "gtm_ctl start" right after
> gtminit because users have to resolve mandatory settings such as
> "host", "port", "gym_host" and "gtm_port" (the latter two are just for
> gtm_proxy.conf).
>
> So it seems to me that gtminit just creates work director and copies
> sample gtm.conf or gtm_proxy.conf and it seems to me that gtminit
> doesn't help much.   Are there any other things which gtminit helps?
>
> Regards;
> ----------
> Koichi Suzuki
>
>
>
> 2012/2/10 Michael Paquier <mic...@gm...>:
> > Hi all,
> >
> > Please find a patch to support a new module called initgtm.
> > It is based on initdb, and initializes the data folder of GTM,
> particularly
> > useful to avoid copying configuration files directly from share folder
> for
> > GTM and GTM-Proxy, in case user has no access to the install repository.
> > The spec of the module is as follows:
> > $ initgtm --help
> > initgtm initializes a GTM for Postgres-XC database cluster.
> >
> > Usage:
> >   initgtm [NODE-TYPE] [OPTION]... [DATADIR]
> >
> > Options:
> >  [-D, --pgdata=]DATADIR     location for this GTM node
> >  [-Z]NODE-TYPE              can be "gtm" or "gtm_proxy"
> > Less commonly used options:
> >   -d, --debug               generate lots of debugging output
> >   -n, --noclean             do not clean up after errors
> >   -s, --show                show internal settings
> >
> > Other options:
> >   -?, --help                show this help, then exit
> >   -V, --version             output version information, then exit
> >
> > Comments are welcome.
> > --
> > Michael Paquier
> > http://michael.otacoo.com
> >
> >
> ------------------------------------------------------------------------------
> > Virtualization & Cloud Management Using Capacity Planning
> > Cloud computing makes use of virtualization - but cloud computing
> > also focuses on allowing computing to be delivered as a service.
> > http://www.accelacomm.com/jaw/sfnl/114/51521223/
> > _______________________________________________
> > Postgres-xc-developers mailing list
> > Pos...@li...
> > https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
> >
>
>
> ------------------------------------------------------------------------------
> Virtualization & Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
> _______________________________________________
> Postgres-xc-developers mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>



-- 
Best Wishes,
Ashutosh Bapat
EntepriseDB Corporation
The Enterprise Postgres Company

Hi,

Gtm.conf and gtm_proxy.conf file cannot be used as is.   This is
different from vanilla PG's initdb.   In vanilla PG, we can issue
"pg_ctl start" just after initdb, if people are happy with default
settings.

In this implementation, we cannot issue "gtm_ctl start" right after
gtminit because users have to resolve mandatory settings such as
"host", "port", "gym_host" and "gtm_port" (the latter two are just for
gtm_proxy.conf).

So it seems to me that gtminit just creates work director and copies
sample gtm.conf or gtm_proxy.conf and it seems to me that gtminit
doesn't help much.   Are there any other things which gtminit helps?

Regards;
----------
Koichi Suzuki



2012/2/10 Michael Paquier <mic...@gm...>:
> Hi all,
>
> Please find a patch to support a new module called initgtm.
> It is based on initdb, and initializes the data folder of GTM, particularly
> useful to avoid copying configuration files directly from share folder for
> GTM and GTM-Proxy, in case user has no access to the install repository.
> The spec of the module is as follows:
> $ initgtm --help
> initgtm initializes a GTM for Postgres-XC database cluster.
>
> Usage:
>   initgtm [NODE-TYPE] [OPTION]... [DATADIR]
>
> Options:
>  [-D, --pgdata=]DATADIR     location for this GTM node
>  [-Z]NODE-TYPE              can be "gtm" or "gtm_proxy"
> Less commonly used options:
>   -d, --debug               generate lots of debugging output
>   -n, --noclean             do not clean up after errors
>   -s, --show                show internal settings
>
> Other options:
>   -?, --help                show this help, then exit
>   -V, --version             output version information, then exit
>
> Comments are welcome.
> --
> Michael Paquier
> http://michael.otacoo.com
>
> ------------------------------------------------------------------------------
> Virtualization & Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
> _______________________________________________
> Postgres-xc-developers mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>

Hi all,

Please find a patch to support a new module called initgtm.
It is based on initdb, and initializes the data folder of GTM, particularly
useful to avoid copying configuration files directly from share folder for
GTM and GTM-Proxy, in case user has no access to the install repository.
The spec of the module is as follows:
$ initgtm --help
initgtm initializes a GTM for Postgres-XC database cluster.

Usage:
  initgtm [NODE-TYPE] [OPTION]... [DATADIR]

Options:
 [-D, --pgdata=]DATADIR     location for this GTM node
 [-Z]NODE-TYPE              can be "gtm" or "gtm_proxy"
Less commonly used options:
  -d, --debug               generate lots of debugging output
  -n, --noclean             do not clean up after errors
  -s, --show                show internal settings

Other options:
  -?, --help                show this help, then exit
  -V, --version             output version information, then exit

Comments are welcome.
-- 
Michael Paquier
http://michael.otacoo.com