zero downtime upgrade to 1.8.0

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

zero downtime upgrade to 1.8.0

Mark Bean
I attempted to upgrade a 3-node Cluster from 1.6.0 to 1.8.0. The process I
used was to configure 1.8.0 on all Nodes. Then, one Node at a time,
shutdown the 1.6.0 and startup the 1.8.0 version. For the first Node, I
copied the flow.xml.gz from the previous 1.6.0 installation. (The
authorizations.xml and users.xml was already copied over and no user, user
group or access policy changes took place during the transition.) The
remaining nodes obtained the flow.xml.gz from the cluster on startup.

The process seemed to be successful. Items in the flow remained during the
transition because I referenced the same repositories.

However, I am not able to access the Global Menu nor Provenance. Note the
provenance repo was upgraded from PersistentProvenanceRepository to
WriteAheadProvenanceRepository in the process.

Can someone explain why these features on not available in the upgraded
cluster?

Thanks,
Mark
Reply | Threaded
Open this post in threaded view
|

Re: zero downtime upgrade to 1.8.0

Joe Witt
Mark,

I *think* there are new authorizations you'd need to give yourself in
that case.  Have you already looked there?

And finally, hard refresh to clear cache is also a possible path

Thanks
On Fri, Nov 2, 2018 at 12:26 PM Mark Bean <[hidden email]> wrote:

>
> I attempted to upgrade a 3-node Cluster from 1.6.0 to 1.8.0. The process I
> used was to configure 1.8.0 on all Nodes. Then, one Node at a time,
> shutdown the 1.6.0 and startup the 1.8.0 version. For the first Node, I
> copied the flow.xml.gz from the previous 1.6.0 installation. (The
> authorizations.xml and users.xml was already copied over and no user, user
> group or access policy changes took place during the transition.) The
> remaining nodes obtained the flow.xml.gz from the cluster on startup.
>
> The process seemed to be successful. Items in the flow remained during the
> transition because I referenced the same repositories.
>
> However, I am not able to access the Global Menu nor Provenance. Note the
> provenance repo was upgraded from PersistentProvenanceRepository to
> WriteAheadProvenanceRepository in the process.
>
> Can someone explain why these features on not available in the upgraded
> cluster?
>
> Thanks,
> Mark
Reply | Threaded
Open this post in threaded view
|

Re: zero downtime upgrade to 1.8.0

Mark Bean
Joe,

The new authorizations should not impact the availability of the Global
Menu. And, your second suggestion was spot on. I did a hard refresh and
that cleared it up.

Thanks!
-Mark

On Fri, Nov 2, 2018 at 12:57 PM Joe Witt <[hidden email]> wrote:

> Mark,
>
> I *think* there are new authorizations you'd need to give yourself in
> that case.  Have you already looked there?
>
> And finally, hard refresh to clear cache is also a possible path
>
> Thanks
> On Fri, Nov 2, 2018 at 12:26 PM Mark Bean <[hidden email]> wrote:
> >
> > I attempted to upgrade a 3-node Cluster from 1.6.0 to 1.8.0. The process
> I
> > used was to configure 1.8.0 on all Nodes. Then, one Node at a time,
> > shutdown the 1.6.0 and startup the 1.8.0 version. For the first Node, I
> > copied the flow.xml.gz from the previous 1.6.0 installation. (The
> > authorizations.xml and users.xml was already copied over and no user,
> user
> > group or access policy changes took place during the transition.) The
> > remaining nodes obtained the flow.xml.gz from the cluster on startup.
> >
> > The process seemed to be successful. Items in the flow remained during
> the
> > transition because I referenced the same repositories.
> >
> > However, I am not able to access the Global Menu nor Provenance. Note the
> > provenance repo was upgraded from PersistentProvenanceRepository to
> > WriteAheadProvenanceRepository in the process.
> >
> > Can someone explain why these features on not available in the upgraded
> > cluster?
> >
> > Thanks,
> > Mark
>