You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by Matteo Bertozzi <th...@gmail.com> on 2016/01/26 19:58:38 UTC
HBase 1.x to 2.0 upgrade goals?
Hey,
what are your goals/hopes for the 1.x to 2.0 migration?
did we decided we can't have another cluster shutdown -> upgrade -> restart?
or if we have replication working between 1.x and 2.x this is a valid
option?
one of the goal of the new Assignment is to be able to handle major
migrations by knowing the state of the cluster in terms of "which version
is each region server running" and with the ability to schedule region
migration for fs format changes and similar "major" changes.
(this requires an upgrade performed, Masters first and then RSs. this is
what I always suggested to people but I never checked if this is what we
are suggesting)
also, the above may require (still looking into it) some new calls added to
the 1.x line.
is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
or at least is it acceptable to force people to move to the latest 1.x.y
(e.g. 1.1.232)?
did we decided that before going to 2.0 you have to be on a 1.x? meaning no
direct upgrade from 0.98 to 2.0?
and for how long should we keep upgrade/compatibility code in 2.x?
can we force people to upgrade to 2.0 and then to the next 2.x?
thoughts?
Re: HBase 1.x to 2.0 upgrade goals?
Posted by Enis Söztutar <en...@gmail.com>.
On Tue, Jan 26, 2016 at 11:33 AM, Matteo Bertozzi <th...@gmail.com>
wrote:
> On Tue, Jan 26, 2016 at 11:15 AM, Enis Söztutar <en...@apache.org> wrote:
>
> > Yep, my feeling is that we should make this rolling upgradable. Having
> > coordination for master to be upgraded first then region servers is fine,
> > but may end up to be problematic in practice. I am not sure how we can
> > enforce this (new regionservers cannot talk to old masters for example,
> or
> > cannot come up due to fs changes?).
> >
>
> I think the upgrade deploy order is crucial here, to be able to have
> rolling upgrades.
> the trick is to rely on the master as the "migration coordinator" since it
> is the only one aware of the full cluster.
> but is basically what we do now by hand if we want to enable new features
> which is more or less like a double rolling restart.
> - upgrade one of the master (nothing changes in format or rpc calls)
> - upgrade the backup masters (nothing changes in format or rpc calls)
> - start upgrading the RSs (nothing changes in format or rpc calls)
> - all the RSs are on the new version
> - we can start triggering migrations as if it is simple region moves from
> the balancer
>
For things like the new assignment manager for example, I think we can make
a totally different and non-compatible implementation using a technique
like you suggest.
We can have the new master come up, at which time, it will look for its own
version and the percentage of regionservers having a version at most his
own. It will wait until all of the regionservers reported are upgraded. If
all of the reported regionservers are upgraded, the master then switches to
the new implementation, and starts rejecting regionservers to join the
cluster with an older version. This is also persisted in zk, so that it
happens once.
>
>
> > > > if we have replication working between 1.x and 2.x
> > >
> > > This is required I think. Stranded users on 0.94 went so far as to hack
> > new
> > > replication endpoints to get it working between 0.94 and 0.96+.
> > >
> >
> > +1. The RPC wire format is not changing in 2.0 is it?
> >
>
> yeah, there should be no wire changes. new methods added but as usual.
> and I'm pretty sure replication-v2 was described as something that run
> parallel to v1 (or something like that)
>
>
> > > > is it acceptable to force people to move to the latest 1.x (e.g.
> > 1.5.x)?
> > >
> > > If this would be the only way to upgrade without downtime, then yes,
> we'd
> > > take it.
> > >
> >
> > Agreed. It is acceptable to require 1.x latest. It should be a last
> resort
> > though.
> >
> > 0.98 -> 2.0 will depend on whether latest 1.x is required or not I think.
>
>
> The 1.x changes should be mainly to be able to have less compact code in
> 2.0
> we can probably do without (still looking into that), but one thing for
> sure is that
> the changes are like new rpc calls and similar, so they can be easily added
> to a 1.1.y or even 0.98.y
> if we don't want to force people to go to the latest 1.x
>
Re: HBase 1.x to 2.0 upgrade goals?
Posted by Matteo Bertozzi <th...@gmail.com>.
On Tue, Jan 26, 2016 at 11:15 AM, Enis Söztutar <en...@apache.org> wrote:
> Yep, my feeling is that we should make this rolling upgradable. Having
> coordination for master to be upgraded first then region servers is fine,
> but may end up to be problematic in practice. I am not sure how we can
> enforce this (new regionservers cannot talk to old masters for example, or
> cannot come up due to fs changes?).
>
I think the upgrade deploy order is crucial here, to be able to have
rolling upgrades.
the trick is to rely on the master as the "migration coordinator" since it
is the only one aware of the full cluster.
but is basically what we do now by hand if we want to enable new features
which is more or less like a double rolling restart.
- upgrade one of the master (nothing changes in format or rpc calls)
- upgrade the backup masters (nothing changes in format or rpc calls)
- start upgrading the RSs (nothing changes in format or rpc calls)
- all the RSs are on the new version
- we can start triggering migrations as if it is simple region moves from
the balancer
> > > if we have replication working between 1.x and 2.x
> >
> > This is required I think. Stranded users on 0.94 went so far as to hack
> new
> > replication endpoints to get it working between 0.94 and 0.96+.
> >
>
> +1. The RPC wire format is not changing in 2.0 is it?
>
yeah, there should be no wire changes. new methods added but as usual.
and I'm pretty sure replication-v2 was described as something that run
parallel to v1 (or something like that)
> > > is it acceptable to force people to move to the latest 1.x (e.g.
> 1.5.x)?
> >
> > If this would be the only way to upgrade without downtime, then yes, we'd
> > take it.
> >
>
> Agreed. It is acceptable to require 1.x latest. It should be a last resort
> though.
>
> 0.98 -> 2.0 will depend on whether latest 1.x is required or not I think.
The 1.x changes should be mainly to be able to have less compact code in 2.0
we can probably do without (still looking into that), but one thing for
sure is that
the changes are like new rpc calls and similar, so they can be easily added
to a 1.1.y or even 0.98.y
if we don't want to force people to go to the latest 1.x
Re: HBase 1.x to 2.0 upgrade goals?
Posted by Enis Söztutar <en...@apache.org>.
On Tue, Jan 26, 2016 at 11:04 AM, Andrew Purtell <ap...@apache.org>
wrote:
> From the perspective of an operator:
>
> > we can't have another cluster shutdown -> upgrade -> restart?
>
> Rolling upgradeable, please. If it is useful to make a distinction between
> regionserver and master roles, and say all of one must be upgraded before
> the other, this is manageable.
>
Yep, my feeling is that we should make this rolling upgradable. Having
coordination for master to be upgraded first then region servers is fine,
but may end up to be problematic in practice. I am not sure how we can
enforce this (new regionservers cannot talk to old masters for example, or
cannot come up due to fs changes?).
>
> > if we have replication working between 1.x and 2.x
>
> This is required I think. Stranded users on 0.94 went so far as to hack new
> replication endpoints to get it working between 0.94 and 0.96+.
>
+1. The RPC wire format is not changing in 2.0 is it?
>
> > is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
>
> If this would be the only way to upgrade without downtime, then yes, we'd
> take it.
>
Agreed. It is acceptable to require 1.x latest. It should be a last resort
though.
0.98 -> 2.0 will depend on whether latest 1.x is required or not I think.
>
>
> On Tue, Jan 26, 2016 at 10:58 AM, Matteo Bertozzi <theo.bertozzi@gmail.com
> >
> wrote:
>
> > Hey,
> >
> > what are your goals/hopes for the 1.x to 2.0 migration?
> >
> > did we decided we can't have another cluster shutdown -> upgrade ->
> > restart?
> > or if we have replication working between 1.x and 2.x this is a valid
> > option?
> >
> > one of the goal of the new Assignment is to be able to handle major
> > migrations by knowing the state of the cluster in terms of "which version
> > is each region server running" and with the ability to schedule region
> > migration for fs format changes and similar "major" changes.
> > (this requires an upgrade performed, Masters first and then RSs. this is
> > what I always suggested to people but I never checked if this is what we
> > are suggesting)
> >
> > also, the above may require (still looking into it) some new calls added
> to
> > the 1.x line.
> > is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
> > or at least is it acceptable to force people to move to the latest 1.x.y
> > (e.g. 1.1.232)?
> >
> > did we decided that before going to 2.0 you have to be on a 1.x? meaning
> no
> > direct upgrade from 0.98 to 2.0?
> >
> > and for how long should we keep upgrade/compatibility code in 2.x?
> > can we force people to upgrade to 2.0 and then to the next 2.x?
> >
> > thoughts?
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
>
Re: HBase 1.x to 2.0 upgrade goals?
Posted by Andrew Purtell <ap...@apache.org>.
>From the perspective of an operator:
> we can't have another cluster shutdown -> upgrade -> restart?
Rolling upgradeable, please. If it is useful to make a distinction between
regionserver and master roles, and say all of one must be upgraded before
the other, this is manageable.
> if we have replication working between 1.x and 2.x
This is required I think. Stranded users on 0.94 went so far as to hack new
replication endpoints to get it working between 0.94 and 0.96+.
> is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
If this would be the only way to upgrade without downtime, then yes, we'd
take it.
On Tue, Jan 26, 2016 at 10:58 AM, Matteo Bertozzi <th...@gmail.com>
wrote:
> Hey,
>
> what are your goals/hopes for the 1.x to 2.0 migration?
>
> did we decided we can't have another cluster shutdown -> upgrade ->
> restart?
> or if we have replication working between 1.x and 2.x this is a valid
> option?
>
> one of the goal of the new Assignment is to be able to handle major
> migrations by knowing the state of the cluster in terms of "which version
> is each region server running" and with the ability to schedule region
> migration for fs format changes and similar "major" changes.
> (this requires an upgrade performed, Masters first and then RSs. this is
> what I always suggested to people but I never checked if this is what we
> are suggesting)
>
> also, the above may require (still looking into it) some new calls added to
> the 1.x line.
> is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
> or at least is it acceptable to force people to move to the latest 1.x.y
> (e.g. 1.1.232)?
>
> did we decided that before going to 2.0 you have to be on a 1.x? meaning no
> direct upgrade from 0.98 to 2.0?
>
> and for how long should we keep upgrade/compatibility code in 2.x?
> can we force people to upgrade to 2.0 and then to the next 2.x?
>
> thoughts?
>
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: HBase 1.x to 2.0 upgrade goals?
Posted by Stack <st...@duboce.net>.
On Tue, Jan 26, 2016 at 2:58 PM, Matteo Bertozzi <th...@gmail.com>
wrote:
> Hey,
>
> what are your goals/hopes for the 1.x to 2.0 migration?
>
> did we decided we can't have another cluster shutdown -> upgrade ->
> restart?
> or if we have replication working between 1.x and 2.x this is a valid
> option?
>
>
Has to be rolling upgradeable I'd say. Our long-time users will never
forgive us if we require another stop-the-world.
> one of the goal of the new Assignment is to be able to handle major
> migrations by knowing the state of the cluster in terms of "which version
> is each region server running" and with the ability to schedule region
> migration for fs format changes and similar "major" changes.
> (this requires an upgrade performed, Masters first and then RSs. this is
> what I always suggested to people but I never checked if this is what we
> are suggesting)
>
>
Refguide 'suggests' this
http://hbase.apache.org/book.html#hbase.rolling.upgrade We should harden it
to so master-first is required.
> also, the above may require (still looking into it) some new calls added to
> the 1.x line.
> is it acceptable to force people to move to the latest 1.x (e.g. 1.5.x)?
> or at least is it acceptable to force people to move to the latest 1.x.y
> (e.g. 1.1.232)?
>
>
I think this is acceptable.... If new migration APIs are being added,
should be a 1.5 rather than a 1.1.30000000000000
> did we decided that before going to 2.0 you have to be on a 1.x? meaning no
> direct upgrade from 0.98 to 2.0?
>
>
0.98 to 2.0 would be tough given nature of the changes in 2.0 and then the
need of new API to help migration. Would be sweet though if a 0.98->2.0
were possible... but migration is already hard enough to dev and test.. The
less migration path permutations supported, the better.
> and for how long should we keep upgrade/compatibility code in 2.x?
> can we force people to upgrade to 2.0 and then to the next 2.x?
>
>
I'd say its fine to purge migration code promptly after 2.0. Better to be
clear on what paths are migratable.
St.Ack
> thoughts?
>