Hi,

I think we should remove the page from the refguide.

I won't remove that tool from Lucene's JARs as it is a good example vor the

Actually there are many othe rthings some customers of myself wanted to upgrade. I you have one, lead them to me. I helped a customer which was unable to reindex and stuck on 3.x to go from Lucene 3 to Lucene 6 with also converting all NumericFields to PointFields and adding docValues in the process too. But this is not as easy as a simple IndexUpgrader tool. If it would not be so hard to setup (you need several versions installed in parallel), I would open it up on github.

The tool does the following:

 * adds DocValues for all fields using the Uninverter (still available
   in Lucene 6.x). It basically does the same like Solr, it just uses
   addIndexes(CodecReader) with a Custom CodecReader and patch of
   FieldInfos.
 * converts NumericField to PointField. Same here, it just uses
   addIndexes(CodecReader), but the CodecReader patches Fieldinfos and
   lucily it is able to use the termsenum of the legacy numeric field
   to extract the numeric values and then during merging it feeds the
   numeric values (which are luckily also sorted correctly) to the
   BKDWriter. The BKDWriter then does the rest and expands the tree
   automatically :-)

So basically some people want to convert old indexes, but the plain IndexUpgrader is of little use. The only thing it does is to upgrade all segments with older versions to actual codec version. This would happen during normal use, too. The main reason why it existed was for the migration 3 -> 4 where the terms enum has to change the order during iteration (UTF-16  of Lucene 3 has a different order than UTF-8 starting with Lucene 4). This slowed down searches, so people wanted to merge all their segments to newest version.

Uwe

Am 04.06.2024 um 01:13 schrieb Jan Høydahl:
I used it back in the days when you could migrate from v3->4->5->6. It solved
the issue that Solr 6 could only read a Lucene 6 or Lucene 5 index, but after 
the
sequence of upgrades you'd get there. I even wrote a wrapper to automate it all 
at
https://github.com/cominvent/solr-tools/tree/master/upgradeindex

However, now that Lucene 9 refuses to work with any index created before Lucene 
8,
I agree with Jason that I cannot see any utility for the upgrade tool anymore. 
Since,
if you have a Solr 6 index, it won't work in Solr 9 even if you used the 
upgrader.
And if you already have a Lucene 8 index, no need to upgrade by hand, just start
Solr 9 on top of the Lucene 8 index and it will be upgraded.

+1 to remove mention from ref-guide.

PS: If the upgrader tool was able to patch the original-index-version from the 
binary
index (and user understands the risks), then there could be some utility. But 
I'm not
aware of such an option.

Jan

3. juni 2024 kl. 21:20 skrev Gus Heck<gus.h...@gmail.com>:

I've fielded many questions on this from clients. Folks who have managed
databases expect to be able to upgrade the data serially across versions
and such, so these questions come up alot with organizations early in their
journey with search. Essentially, I tell them that it's a stop gap tool for
use if there's a serious security issue and you really need to move up one
version to get away from the security issue, but otherwise the correct
procedure is to re-index. (And this is sometimes when I find out if they've
really planned for reindexing or not). A down-side of the existence of this
tool is that its presence in the ref guide allows people to prolong the
period where they confuse managing a search index with managing a database.
That said, I think it may have a place to help folks who got going without
knowing enough about search engines to extend the period they have to dig
themselves out of bad situations (by allowing an upgrade while they figure
out how they are going to handle re-indexing more data than they previously
planned on indexing all at once). So far every one of my clients has taken
my advice and simply re-indexed (though not always without grumbling!), so
I have to admit I've not actually seen it used, but in theory it makes some
sense.

-Gus

On Mon, Jun 3, 2024 at 2:20 PM David Smiley<dsmi...@apache.org>  wrote:

FWIW, in my experience I've never run this tool (nor colleagues) at
any stage in my career that I can remember.  For one reason, all the
systems could re-index if they needed to.
It may be best to remove this information, as it could introduce more
confusion than it helps.

On Mon, Jun 3, 2024 at 1:34 PM Jason Gerlowski<gerlowsk...@gmail.com>
wrote:
Hey all,

I was poking around the ref-guide a bit recently and noticed our page
on the "IndexUpgraderTool" that Lucene produces. [1]

AFAICT, the page doesn't hint at when/why a user might want to use the
IndexUpgraderTool.  Maybe at one point the tool might've been
preferred to loading the index in an upgraded Solr version, but I
haven't heard of anyone doing that in the last few years.

Is this something we expect users to still do?  If so, for what
usecase?  And if not - should we drop it from the ref-guide - it seems
like it might confuse folks since it's not actually needed to upgrade
Solr versions...

Best,

Jason

[1]
https://solr.apache.org/guide/solr/latest/deployment-guide/indexupgrader-tool.html
---------------------------------------------------------------------
To unsubscribe, e-mail:dev-unsubscr...@solr.apache.org
For additional commands, e-mail:dev-h...@solr.apache.org

---------------------------------------------------------------------
To unsubscribe, e-mail:dev-unsubscr...@solr.apache.org
For additional commands, e-mail:dev-h...@solr.apache.org


--
http://www.needhamsoftware.com  (work)
https://a.co/d/b2sZLD9  (my fantasy fiction book)

--
Uwe Schindler
Achterdiek 19, D-28357 Bremen
https://www.thetaphi.de
eMail:u...@thetaphi.de

Reply via email to