Make sure you use fully qualified domain names everywhere, and that they are routable everywhere. If you start Solr with -h my-short-name then it will be persisted in zk state and will not be routable from all other solr hosts.
Jan Høydahl > 25. apr. 2024 kl. 13:34 skrev Ritesh Kumar (Accenture International Limite) > <v-kur...@microsoft.com.invalid>: > > Hello, > > One additional observation, I do not face the issue if I use IP addresses for > the solr nodes but I observe this issue as soon as I modify solr service to > use hostname instead of the IP address. > > start -f -h XX.XX.XX.XX -p 8983 -cloud -Dsolr.ssl.checkPeerName=false -m 8g > -z > "XX.X.X.XX:2181,YY.Y.Y.YY:2181,ZZ.Z.Z.ZZ:2181,AA.A.A.AA:2181,BB.B.B.BB:2181" > > start -f -h HOSTNAME -p 8983 -cloud -Dsolr.ssl.checkPeerName=false -m 8g -z > "XX.X.X.XX:2181,YY.Y.Y.YY:2181,ZZ.Z.Z.ZZ:2181,AA.A.A.AA:2181,BB.B.B.BB:2181" > > > Thanks > > Ritesh > > ________________________________ > From: Jan H?ydahl <jan....@cominvent.com> > Sent: Wednesday, April 24, 2024 02:21 PM > To: users@solr.apache.org <users@solr.apache.org> > Subject: Re: [EXTERNAL] SOLR Shard nodes are unhealthy > > Hi, > > Indeed, Solr 8.11 used zk client v3.6.2, but I believe it will also work with > newer 3.x versions, you can test this locally on your laptop first to be > sure. > > Note also that network resolution should work between all nodes on the FQDN. > So if your ZK1 node is called ZK1.region3.example.com and is on ip 10.0.1.1 > then it should be routable on that FQDN address from all solr and zk nodes, > and a reverse lookup on IP should also return the same host name. > > Jan > >> 24. apr. 2024 kl. 10:38 skrev Ritesh Kumar (Accenture International Limite) >> <v-kur...@microsoft.com.INVALID>: >> >> Hi Jan, >> >> Your assumptions are correct as those are setup correctly. I will provide >> the logs. >> >> Meanwhile I can give a try with SOLR 8.11.3 and see how it behaves. Moreover >> just wanted to check one thing: >> >> SOLR 8.11 document says "Solr currently uses Apache ZooKeeper v3.6.2." >> Setting Up an External ZooKeeper Ensemble | Apache Solr Reference Guide >> 8.11<https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fsolr.apache.org%2Fguide%2F8_11%2Fsetting-up-an-external-zookeeper-ensemble.html&data=05%7C02%7Cv-kurite%40microsoft.com%7Ca075331631884b3cfb1808dc643bc41a%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638495455093288541%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=2bN%2B4oq9MJcmNv315lnR4Op9OvPG4yP0cF%2FQDye9%2Ffs%3D&reserved=0<https://solr.apache.org/guide/8_11/setting-up-an-external-zookeeper-ensemble.html>> >> >> However looking at Apache >> ZooKeeper<https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fzookeeper.apache.org%2Freleases.html&data=05%7C02%7Cv-kurite%40microsoft.com%7Ca075331631884b3cfb1808dc643bc41a%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638495455093299320%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=78iVlW3%2BRilLuGrFr%2FzqdnYYRdDvM5H891djm4ZFI%2B0%3D&reserved=0<https://zookeeper.apache.org/releases.html>>, >> it says "Apache ZooKeeper 3.9.2 is our current release, and 3.8.4 our >> latest stable release." >> >> Hence i wanted to check if i can install zookeeper 3.8.4 along with solr >> 8.11.3 or i have to use ZooKeeper v3.6.2 only. >> >> >> Thanks >> >> Ritesh >> >> ________________________________ >> From: Jan H?ydahl <jan....@cominvent.com> >> Sent: Wednesday, April 24, 2024 01:41 PM >> To: users@solr.apache.org <users@solr.apache.org> >> Subject: [EXTERNAL] Re: SOLR Shard nodes are unhealthy >> >> Hi, >> >> First, you should use the latest 8.11.3 which has a lot of fixes. >> >> Your setup could work if the regions are connected with really low latency >> and high bandwidth. >> >> Your description suggests that your Solr-ZK connection is not as it should. >> >> A few assumptions for this to work: >> * You ave installed ZK correctly as one five-node ensemble, and validated it >> works flawlessly >> * All your Solr nodes lists all five ZK nodes in ZK_HOST, e.g. >> ZK_HOST=zk1:2181,zk2:2182,zk3:2181,zk4:2181,zk5:2181 >> >> You'll have to provide more information regarding your exact configuration >> and/or error log messages to get further on this. >> >> Jan >> >>>> 24. apr. 2024 kl. 04:32 skrev Ritesh Kumar (Accenture International >>>> Limite) <v-kur...@microsoft.com.INVALID>: >>> >>> Hello All, >>> >>> I have setup SOLR-ZK ensemble (Windows) with 6 SOLR VMs across two regions >>> and 5 ZK VMs across three regions. >>> >>> SOLR Version - 8.11.2 >>> ZK Version - 3.6.2 >>> >>> Region1: >>> SOLR1 >>> SOLR2 >>> SOLR3 >>> ZK2 >>> ZK3 >>> >>> Region2: >>> SOLR4 >>> SOLR5 >>> SOLR6 >>> ZK4 >>> ZK5 >>> >>> Region3: >>> ZK1 >>> >>> I am facing issue after restarting ZK and SOLR VMs . The SOLR UI is showing >>> all the 6 SOLR nodes and 5 ZK nodes are active and with leader. >>> >>> All the 3 Solr nodes (1,2,3) of each shard is in recovery and later down >>> and remaining 3 solr nodes (4,5,6) are healthy. Now if i restart solr nodes >>> 4,5,6 they go to recovery/down and nodes 1,2,3 become healthy - but all 6 >>> nodes are never becoming healthy together. I have tried restarting Solr and >>> ZK multiple times. >>> >>> Please let me know if anyone has faced this issue or has suggestions for a >>> fix. >>> >>> >>> >>> >>> Note: I cannot upgrade to latest SOLR version as this is recommened by >>> Sitecore. >>> >>> Thanks >>> Ritesh >> >
