Hello,

My old cluster is running 0.90.4 and the new cluster is running 0.92.1

I managed to copy the data from old cluster to new cluster using

hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp

I even copied it to the newcluster :/hbase

I am not sure what I need to do next to have hbase recognize the data

--
--- Get your facts first, then you can distort them as you please.--

When you start the new cluster with 0.92.1, hbase would migrate your data to 0.92 format.

Thanks

On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:

> Hello,
>
> My old cluster is running 0.90.4 and the new cluster is running 0.92.1
>
> I managed to copy the data from old cluster to new cluster using
>
> hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
>
> I even copied it to the newcluster :/hbase
>
> I am not sure what I need to do next to have hbase recognize the data
>
> --
> --- Get your facts first, then you can distort them as you please.--

Do I need to restart the cluster? My 0.92.1 already has several hbase
tables.

On Sat, Nov 17, 2012 at 9:54 AM, <[EMAIL PROTECTED]> wrote:

> When you start the new cluster with 0.92.1, hbase would migrate your data
> to 0.92 format.
>
> Thanks
>
>
>
> On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:
>
> > Hello,
> >
> > My old cluster is running 0.90.4 and the new cluster is running 0.92.1
> >
> > I managed to copy the data from old cluster to new cluster using
> >
> > hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
> >
> > I even copied it to the newcluster :/hbase
> >
> > I am not sure what I need to do next to have hbase recognize the data
> >
> > --
> > --- Get your facts first, then you can distort them as you please.--
>

--
--- Get your facts first, then you can distort them as you please.--

So the .META. table in your 0.92.1 cluster doesn't have rows for the copied
tables.

Looks like you need:
https://issues.apache.org/jira/browse/HBASE-3271
Allow .META. table to be exported

But it is in 0.94.2 and above.

Cheers

On Sat, Nov 17, 2012 at 8:41 AM, Rita <[EMAIL PROTECTED]> wrote:

> Do I need to restart the cluster? My 0.92.1 already has several hbase
> tables.
>
> On Sat, Nov 17, 2012 at 9:54 AM, <[EMAIL PROTECTED]> wrote:
>
> > When you start the new cluster with 0.92.1, hbase would migrate your data
> > to 0.92 format.
> >
> > Thanks
> >
> >
> >
> > On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:
> >
> > > Hello,
> > >
> > > My old cluster is running 0.90.4 and the new cluster is running 0.92.1
> > >
> > > I managed to copy the data from old cluster to new cluster using
> > >
> > > hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
> > >
> > > I even copied it to the newcluster :/hbase
> > >
> > > I am not sure what I need to do next to have hbase recognize the data
> > >
> > > --
> > > --- Get your facts first, then you can distort them as you please.--
> >
>
>
>
> --
> --- Get your facts first, then you can distort them as you please.--
>

Assuming the tables from 0.90.4 cluster are absent in 0.92.1 cluster, you
can (I haven't tried this):
1. sideline the data you just copied from 0.90.4 cluster
2. create table(s) in 0.92.1 cluster with same schema as those in 0.90.4
cluster - you need to presplit new table with the exact same region
boundaries as the old table
3. move back the sidelined data into corresponding regions

Thanks

On Sat, Nov 17, 2012 at 8:47 AM, Ted Yu <[EMAIL PROTECTED]> wrote:

> So the .META. table in your 0.92.1 cluster doesn't have rows for the
> copied tables.
>
> Looks like you need:
> https://issues.apache.org/jira/browse/HBASE-3271
> Allow .META. table to be exported
>
> But it is in 0.94.2 and above.
>
> Cheers
>
>
> On Sat, Nov 17, 2012 at 8:41 AM, Rita <[EMAIL PROTECTED]> wrote:
>
>> Do I need to restart the cluster? My 0.92.1 already has several hbase
>> tables.
>>
>> On Sat, Nov 17, 2012 at 9:54 AM, <[EMAIL PROTECTED]> wrote:
>>
>> > When you start the new cluster with 0.92.1, hbase would migrate your
>> data
>> > to 0.92 format.
>> >
>> > Thanks
>> >
>> >
>> >
>> > On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:
>> >
>> > > Hello,
>> > >
>> > > My old cluster is running 0.90.4 and the new cluster is running 0.92.1
>> > >
>> > > I managed to copy the data from old cluster to new cluster using
>> > >
>> > > hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
>> > >
>> > > I even copied it to the newcluster :/hbase
>> > >
>> > > I am not sure what I need to do next to have hbase recognize the data
>> > >
>> > > --
>> > > --- Get your facts first, then you can distort them as you please.--
>> >
>>
>>
>>
>> --
>> --- Get your facts first, then you can distort them as you please.--
>>
>
>

Wow. Sounds complicated.

Let me try copying the entire hbase directory over. And then restart the
cluster.

For future release of Hbase the upgrades and table compatibility should be
better supported correct?
On Sat, Nov 17, 2012 at 12:18 PM, Ted Yu <[EMAIL PROTECTED]> wrote:

> Assuming the tables from 0.90.4 cluster are absent in 0.92.1 cluster, you
> can (I haven't tried this):
> 1. sideline the data you just copied from 0.90.4 cluster
> 2. create table(s) in 0.92.1 cluster with same schema as those in 0.90.4
> cluster - you need to presplit new table with the exact same region
> boundaries as the old table
> 3. move back the sidelined data into corresponding regions
>
> Thanks
>
> On Sat, Nov 17, 2012 at 8:47 AM, Ted Yu <[EMAIL PROTECTED]> wrote:
>
> > So the .META. table in your 0.92.1 cluster doesn't have rows for the
> > copied tables.
> >
> > Looks like you need:
> > https://issues.apache.org/jira/browse/HBASE-3271
> > Allow .META. table to be exported
> >
> > But it is in 0.94.2 and above.
> >
> > Cheers
> >
> >
> > On Sat, Nov 17, 2012 at 8:41 AM, Rita <[EMAIL PROTECTED]> wrote:
> >
> >> Do I need to restart the cluster? My 0.92.1 already has several hbase
> >> tables.
> >>
> >> On Sat, Nov 17, 2012 at 9:54 AM, <[EMAIL PROTECTED]> wrote:
> >>
> >> > When you start the new cluster with 0.92.1, hbase would migrate your
> >> data
> >> > to 0.92 format.
> >> >
> >> > Thanks
> >> >
> >> >
> >> >
> >> > On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:
> >> >
> >> > > Hello,
> >> > >
> >> > > My old cluster is running 0.90.4 and the new cluster is running
> 0.92.1
> >> > >
> >> > > I managed to copy the data from old cluster to new cluster using
> >> > >
> >> > > hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
> >> > >
> >> > > I even copied it to the newcluster :/hbase
> >> > >
> >> > > I am not sure what I need to do next to have hbase recognize the
> data
> >> > >
> >> > > --
> >> > > --- Get your facts first, then you can distort them as you please.--
> >> >
> >>
> >>
> >>
> >> --
> >> --- Get your facts first, then you can distort them as you please.--
> >>
> >
> >
>

--
--- Get your facts first, then you can distort them as you please.--

Actually the procedure is not complicated.
Our group recently performed an upgrade of 0.90.x cluster to 0.92 where we
need to merge adjacent regions from the original table into fewer regions
in the new table. This was due to the overly small region size in 0.90.x
cluster.
That required scripting and moving HFiles around.

In terms of compatibility across major releases, we have been consistent:
(one-way) auto-upgrade is provided from one major release to the next major
release.

Cheers

On Sat, Nov 17, 2012 at 10:10 AM, Rita <[EMAIL PROTECTED]> wrote:

> Wow. Sounds complicated.
>
> Let me try copying the entire hbase directory over. And then restart the
> cluster.
>
> For future release of Hbase the upgrades and table compatibility should be
> better supported correct?
>
>
> On Sat, Nov 17, 2012 at 12:18 PM, Ted Yu <[EMAIL PROTECTED]> wrote:
>
> > Assuming the tables from 0.90.4 cluster are absent in 0.92.1 cluster, you
> > can (I haven't tried this):
> > 1. sideline the data you just copied from 0.90.4 cluster
> > 2. create table(s) in 0.92.1 cluster with same schema as those in 0.90.4
> > cluster - you need to presplit new table with the exact same region
> > boundaries as the old table
> > 3. move back the sidelined data into corresponding regions
> >
> > Thanks
> >
> > On Sat, Nov 17, 2012 at 8:47 AM, Ted Yu <[EMAIL PROTECTED]> wrote:
> >
> > > So the .META. table in your 0.92.1 cluster doesn't have rows for the
> > > copied tables.
> > >
> > > Looks like you need:
> > > https://issues.apache.org/jira/browse/HBASE-3271
> > > Allow .META. table to be exported
> > >
> > > But it is in 0.94.2 and above.
> > >
> > > Cheers
> > >
> > >
> > > On Sat, Nov 17, 2012 at 8:41 AM, Rita <[EMAIL PROTECTED]> wrote:
> > >
> > >> Do I need to restart the cluster? My 0.92.1 already has several hbase
> > >> tables.
> > >>
> > >> On Sat, Nov 17, 2012 at 9:54 AM, <[EMAIL PROTECTED]> wrote:
> > >>
> > >> > When you start the new cluster with 0.92.1, hbase would migrate your
> > >> data
> > >> > to 0.92 format.
> > >> >
> > >> > Thanks
> > >> >
> > >> >
> > >> >
> > >> > On Nov 17, 2012, at 6:31 AM, Rita <[EMAIL PROTECTED]> wrote:
> > >> >
> > >> > > Hello,
> > >> > >
> > >> > > My old cluster is running 0.90.4 and the new cluster is running
> > 0.92.1
> > >> > >
> > >> > > I managed to copy the data from old cluster to new cluster using
> > >> > >
> > >> > > hadoop distcp -i hftp://oldcluster:50070/hbase/foo.taq /tmp
> > >> > >
> > >> > > I even copied it to the newcluster :/hbase
> > >> > >
> > >> > > I am not sure what I need to do next to have hbase recognize the
> > data
> > >> > >
> > >> > > --
> > >> > > --- Get your facts first, then you can distort them as you
> please.--
> > >> >
> > >>
> > >>
> > >>
> > >> --
> > >> --- Get your facts first, then you can distort them as you please.--
> > >>
> > >
> > >
> >
>
>
>
> --
> --- Get your facts first, then you can distort them as you please.--
>