it seems I have been slacking on keeping up do date with updates, will it
be possible to upgrade a 0.22.1 cluster to 1.0 or is it better to do it in
two steps i.e. 0.90.x first
it seems I have been slacking on keeping up do date with updates, will it
be possible to upgrade a 0.22.1 cluster to 1.0 or is it better to do it in
two steps i.e. 0.90.x first
Thanks for the seedy reply, I was not thinking of an immediate upgrade
since it needs to be scheduled tested etc, but rather considering skipping
90 since 1.0 is so close to release and doing a one step upgrade instead of
a 2 step so rather than do an upgrade now then again in a few months just
go strait for 1.0 when its final release is out, my main concern is if its
possible to skip 90 in production, of course testing will be done however
it never matches to real word circumstances since replicating a test
environment with 4 nodes and gigabytes of data may in theory be possible
but actual load and results are difficult do.
I suppose in the extreme case we can just do a fresh install of 1.0 and
reindex all the data it just means several hours of 'downtime'
so I guess the real question is which route is easiest as opposed to safest
0.20 -> 0.90 -> 1.0
0.20 -> 1.0
nuke and pave strait to 1.0
GX
On Tuesday, February 4, 2014 7:42:11 AM UTC+2, Mark Walkom wrote:
You will probably want to upgrade to 0.90.0 first.
v1.0 is still RC so if this is production you might want to hold off.
On 4 February 2014 16:21, GX <mail...@gmail.com <javascript:>> wrote:
Hi All
it seems I have been slacking on keeping up do date with updates, will it
be possible to upgrade a 0.22.1 cluster to 1.0 or is it better to do it in
two steps i.e. 0.90.x first
The easiest is to do an upgrade directly to v1.0, but I highly doubt if
that will even work after the upgrade due to the number of changes between
0.2X, 0.90.X and 1.0.0.
And frankly, you'd be insane to consider it if you wanted to keep your data.
If you can export your data to disk, then reimport/reindex it, that might
be an easier option for you.
Thanks for the seedy reply, I was not thinking of an immediate upgrade
since it needs to be scheduled tested etc, but rather considering skipping
90 since 1.0 is so close to release and doing a one step upgrade instead of
a 2 step so rather than do an upgrade now then again in a few months just
go strait for 1.0 when its final release is out, my main concern is if its
possible to skip 90 in production, of course testing will be done however
it never matches to real word circumstances since replicating a test
environment with 4 nodes and gigabytes of data may in theory be possible
but actual load and results are difficult do.
I suppose in the extreme case we can just do a fresh install of 1.0 and
reindex all the data it just means several hours of 'downtime'
so I guess the real question is which route is easiest as opposed to
safest
0.20 -> 0.90 -> 1.0
0.20 -> 1.0
nuke and pave strait to 1.0
GX
On Tuesday, February 4, 2014 7:42:11 AM UTC+2, Mark Walkom wrote:
You will probably want to upgrade to 0.90.0 first.
v1.0 is still RC so if this is production you might want to hold off.
it seems I have been slacking on keeping up do date with updates, will
it be possible to upgrade a 0.22.1 cluster to 1.0 or is it better to do it
in two steps i.e. 0.90.x first
On Tuesday, February 4, 2014 9:04:43 AM UTC+2, Mark Walkom wrote:
The easiest is to do an upgrade directly to v1.0, but I highly doubt if
that will even work after the upgrade due to the number of changes between
0.2X, 0.90.X and 1.0.0.
And frankly, you'd be insane to consider it if you wanted to keep your
data.
If you can export your data to disk, then reimport/reindex it, that might
be an easier option for you.
On 4 February 2014 17:20, GX <mail...@gmail.com <javascript:>> wrote:
Hi Mark
Thanks for the seedy reply, I was not thinking of an immediate upgrade
since it needs to be scheduled tested etc, but rather considering skipping
90 since 1.0 is so close to release and doing a one step upgrade instead of
a 2 step so rather than do an upgrade now then again in a few months just
go strait for 1.0 when its final release is out, my main concern is if its
possible to skip 90 in production, of course testing will be done however
it never matches to real word circumstances since replicating a test
environment with 4 nodes and gigabytes of data may in theory be possible
but actual load and results are difficult do.
I suppose in the extreme case we can just do a fresh install of 1.0 and
reindex all the data it just means several hours of 'downtime'
so I guess the real question is which route is easiest as opposed to
safest
0.20 -> 0.90 -> 1.0
0.20 -> 1.0
nuke and pave strait to 1.0
GX
On Tuesday, February 4, 2014 7:42:11 AM UTC+2, Mark Walkom wrote:
You will probably want to upgrade to 0.90.0 first.
v1.0 is still RC so if this is production you might want to hold off.
it seems I have been slacking on keeping up do date with updates, will
it be possible to upgrade a 0.22.1 cluster to 1.0 or is it better to do it
in two steps i.e. 0.90.x first
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.