You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@solr.apache.org by David Smiley <ds...@apache.org> on 2026/03/28 17:45:29 UTC
Node names vs URI
Internally, SolrCloud distinguishes nodes by a partial URL like
"localhost:8983/solr" (off the top of my head). Yet to actually
communicate with such a node, you need to apply the URL scheme --
"http://" or "https://". Is this serving us well? It seems annoying
to always have to translate and furthermore to not hold references
using a better type like java.net.URI instead of String, which is
basically untyped. Admittedly I haven't had to experience a URL
scheme change... and I don't know if that's even possible with Solr
today.
~ David Smiley
Apache Lucene/Solr Search Developer
http://www.linkedin.com/in/davidwsmiley
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
For additional commands, e-mail: dev-help@solr.apache.org
Re: Node names vs URI
Posted by Gus Heck <gu...@gmail.com>.
There had been cases where the only way to upgrade an existing cluster to
https was to hand-edit the zk state ... I expect this is meant to prevent
the need for hand editing (off the top of my head YMMV)
There's a "todo" lurking in the back of my head to ponder whether we could
get mileage out of having names (or other ids) for nodes and then a single
entirely independent "find nodes at these addresses" file (called
node-addressbook.json or something)
On Sun, Mar 29, 2026 at 10:45 AM Mike Drob <md...@apache.org> wrote:
> I don’t remember the detailed any longer, but Tim Potter worked on this
> problem a few years back and ran into lots of thorns, especially with
> cluster upgrades. I’d look at the discussions from that project for insight
> into your current questions.
>
> Mike
>
> On Sun, Mar 29, 2026 at 5:32 AM Christos Malliaridis <
> c.malliaridis@gmail.com> wrote:
>
> > I see this pattern often, and I also kind of wish it would be
> better-typed.
> > When I am usually migrating code to use URI / URL, I often end up facing
> > issues with serialization, which is probably the main reason String is
> > preferred.
> >
> > But the benefits of using a type like URL when passing a node name around
> > is likely more convenient in many ways, especially the error handling
> that
> > can be avoided to some degree.
> >
> > However, what is the impact of following that route at least for internal
> > code? And since we are storing data into files, what about the
> > serialization and deserialization of that URI? I believe it should not be
> > much of a problem anyway?
> >
> > Christos
> >
> > On Sat, Mar 28, 2026 at 7:45 PM David Smiley <ds...@apache.org> wrote:
> >
> > > Internally, SolrCloud distinguishes nodes by a partial URL like
> > > "localhost:8983/solr" (off the top of my head). Yet to actually
> > > communicate with such a node, you need to apply the URL scheme --
> > > "http://" or "https://". Is this serving us well? It seems annoying
> > > to always have to translate and furthermore to not hold references
> > > using a better type like java.net.URI instead of String, which is
> > > basically untyped. Admittedly I haven't had to experience a URL
> > > scheme change... and I don't know if that's even possible with Solr
> > > today.
> > >
> > > ~ David Smiley
> > > Apache Lucene/Solr Search Developer
> > > http://www.linkedin.com/in/davidwsmiley
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
> > > For additional commands, e-mail: dev-help@solr.apache.org
> > >
> > >
> >
>
--
http://www.needhamsoftware.com (work)
https://a.co/d/b2sZLD9 (my fantasy fiction book)
Re: Node names vs URI
Posted by Mike Drob <md...@apache.org>.
I don’t remember the detailed any longer, but Tim Potter worked on this
problem a few years back and ran into lots of thorns, especially with
cluster upgrades. I’d look at the discussions from that project for insight
into your current questions.
Mike
On Sun, Mar 29, 2026 at 5:32 AM Christos Malliaridis <
c.malliaridis@gmail.com> wrote:
> I see this pattern often, and I also kind of wish it would be better-typed.
> When I am usually migrating code to use URI / URL, I often end up facing
> issues with serialization, which is probably the main reason String is
> preferred.
>
> But the benefits of using a type like URL when passing a node name around
> is likely more convenient in many ways, especially the error handling that
> can be avoided to some degree.
>
> However, what is the impact of following that route at least for internal
> code? And since we are storing data into files, what about the
> serialization and deserialization of that URI? I believe it should not be
> much of a problem anyway?
>
> Christos
>
> On Sat, Mar 28, 2026 at 7:45 PM David Smiley <ds...@apache.org> wrote:
>
> > Internally, SolrCloud distinguishes nodes by a partial URL like
> > "localhost:8983/solr" (off the top of my head). Yet to actually
> > communicate with such a node, you need to apply the URL scheme --
> > "http://" or "https://". Is this serving us well? It seems annoying
> > to always have to translate and furthermore to not hold references
> > using a better type like java.net.URI instead of String, which is
> > basically untyped. Admittedly I haven't had to experience a URL
> > scheme change... and I don't know if that's even possible with Solr
> > today.
> >
> > ~ David Smiley
> > Apache Lucene/Solr Search Developer
> > http://www.linkedin.com/in/davidwsmiley
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
> > For additional commands, e-mail: dev-help@solr.apache.org
> >
> >
>
Re: Node names vs URI
Posted by Christos Malliaridis <c....@gmail.com>.
I see this pattern often, and I also kind of wish it would be better-typed.
When I am usually migrating code to use URI / URL, I often end up facing
issues with serialization, which is probably the main reason String is
preferred.
But the benefits of using a type like URL when passing a node name around
is likely more convenient in many ways, especially the error handling that
can be avoided to some degree.
However, what is the impact of following that route at least for internal
code? And since we are storing data into files, what about the
serialization and deserialization of that URI? I believe it should not be
much of a problem anyway?
Christos
On Sat, Mar 28, 2026 at 7:45 PM David Smiley <ds...@apache.org> wrote:
> Internally, SolrCloud distinguishes nodes by a partial URL like
> "localhost:8983/solr" (off the top of my head). Yet to actually
> communicate with such a node, you need to apply the URL scheme --
> "http://" or "https://". Is this serving us well? It seems annoying
> to always have to translate and furthermore to not hold references
> using a better type like java.net.URI instead of String, which is
> basically untyped. Admittedly I haven't had to experience a URL
> scheme change... and I don't know if that's even possible with Solr
> today.
>
> ~ David Smiley
> Apache Lucene/Solr Search Developer
> http://www.linkedin.com/in/davidwsmiley
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@solr.apache.org
> For additional commands, e-mail: dev-help@solr.apache.org
>
>