I think this might be the problem. I was casting CloudSolrClient object or HttpSolrClient object to SolrClient inside the indexing service and performed add or query operation on this SolrClient object.
I had to cast the client object because the client object may be of any type based on whether the Solr is running in cloud mode or stand-alone mode. For now, I am using the CloudSolrClient. Everything seems to be running fine. I have not even closed the client object after execution. I am still not sure if casting the client object to SolrClient was the issue. I am still looking for an answer where I should be able to run my application in both the Solr modes with the same piece of code barring the client object. On Mon, Jul 2, 2018 at 7:49 PM Erick Erickson <erickerick...@gmail.com> wrote: > Given your other e-mail I suspect you're not closing the client > and creating new ones for every update request. > > You should simply not run out of connections, your client is > most probably incorrect. > > Best, > Erick > > On Mon, Jul 2, 2018 at 3:38 AM, Ritesh Kumar > <ritesh.ku...@hotwaxsystems.com> wrote: > > I could get the live Solr nodes using this piece of code > > > > ZkStateReader zkStateReader = client.getZkStateReader(); > > ClusterState clusterState = zkStateReader.getClusterState(); > > Set<String> liveNodes = clusterState.getLiveNodes(); > > > > This way, I will be able to send a query to one of the live nodes and > > Zookeeper will take care of the rest, but, I was wondering if this is a > > good practice to query from SolrCloud. > > > > What if the Solr node goes down in the middle of bulk indexing. > > > > On Mon, Jul 2, 2018 at 3:37 PM Ritesh Kumar < > ritesh.ku...@hotwaxsystems.com> > > wrote: > > > >> I did use CloudSolrClient to query or index data. I did not have to > check > >> which Solr node is active. The problem I am facing during bulk indexing > is > >> that the Zookeeper runs out of connections resulting in Connection > Timeout > >> error. > >> > >> How can I get to know in advance the active Solr nodes? Any reference > >> would be helpful. > >> > >> Thanks > >> > >> On Mon, Jul 2, 2018 at 2:36 PM Yasufumi Mizoguchi < > yasufumi0...@gmail.com> > >> wrote: > >> > >>> Hi, > >>> > >>> I think ZooKeeper can not notice requests to dead nodes, if you send > >>> requests to Solr nodes directly. > >>> It will be better that asking ZooKeeper which Solr nodes will be > running > >>> before requesting Solr nodes with CloudSolrClient etc... > >>> > >>> Thanks, > >>> Yasufumi > >>> > >>> 2018年7月2日(月) 16:49 Ritesh Kumar <ritesh.ku...@hotwaxsystems.com>: > >>> > >>> > Hello Team, > >>> > > >>> > I have two Solr nodes running in cloud mode. I know that we send > queries > >>> > and updates directly to Solr's collection e.g.http://host: > >>> > port/solr/<your_collection_name>. Any of the Solr nodes can be used. > If > >>> the > >>> > node does not have the collection being queried then the request > will be > >>> > forwarded internally to a Solr instance which has that collection. > >>> > > >>> > But, my question is what happens when the node being queried is > down. I > >>> am > >>> > getting this > >>> > error: Server refused connection at http://localhost:xxxx > >>> > /solr/collectionName. > >>> > > >>> > Does not Zookeeper handle this scenario? > >>> > > >>> > Everything is fine when the node being queried is running. I am able > to > >>> > index and fetch data. > >>> > > >>> > Please, help me. > >>> > > >>> > Best, > >>> > Ritesh Kumar > >>> > > >>> > >> >
