that syntax isn’t isn’t a syntax that Solr recognize intentionally at all.
At very best, the select handler is using the default field (if it’s defined).

Although not having a “q” parameter means all bets are off. On a local copy of 
7.3
I have lying around I get a valid response, but using a MatchNoDocsQuery
(added &debug=query to something similar) so it finds no documents. And no
NPE even with a sharded collection.

What probably happened after you split the shard was that the sub-requests were
being sent slightly differently than the input. When a collection has more than 
one
shard, a subrequest is sent out to one replica of each shard. Apparently the 
syntax
you’re using doesn’t quite survive that process unscathed.

So your single-shard happened to be working with incorrect syntax by chance.

It’d be interesting to see how the query worked, if you have a copy of your
single-shard setup working try adding &debug=query to the url, the return
should have the parsed query which would be informative.

Best,
Erick

> On Sep 3, 2020, at 2:04 PM, Louis <richkingmyst...@gmail.com> wrote:
> 
> Yes, we are sure that this is not typo.
> 
> Actually we did more experiments and found that
> 
> 1) https://hostname:8983/solr/my_collection/select?ids=169455599|1
> 2) https://hostname:8983/solr/my_collection/select?q=id:169455599|1 
> 3) https://hostname:8983/solr/my_collection/get?ids=169455599|1 
> 
> 1) throws NPE, but 2) and 3) is working.
> 
> However 1) is also working with one shard in the same collection which I
> backed up before splitting a shard
> 
> It looks very weird and we are wondering about missing any config. 
> 
> Or, 1) is the type of the syntax Solr discourages to use?
> 
> 
> 
> 
> --
> Sent from: https://lucene.472066.n3.nabble.com/Solr-User-f472068.html

Reply via email to