Apache
/
lucene
mirror of https://github.com/apache/lucene.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

SOLR-6484: SolrCLI's healthcheck action needs to check live nodes as part of reporting the status of a replica 

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1626823 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Timothy Potter 2014-09-22 16:33:56 +00:00
parent 05a55f9786
commit 8a69b1d6cb
2 changed files with 45 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
solr/CHANGES.txt
View File

@ -189,6 +189,9 @@ Bug Fixes
* SOLR-6481: CLUSTERSTATUS should check if the node hosting a replica is live when * SOLR-6481: CLUSTERSTATUS should check if the node hosting a replica is live when
reporting replica status (Timothy Potter) reporting replica status (Timothy Potter)
* SOLR-6484: SolrCLI's healthcheck action needs to check live nodes as part of reporting
the status of a replica (Timothy Potter)
Other Changes Other Changes
---------------------- ----------------------

74
solr/core/src/java/org/apache/solr/util/SolrCLI.java
View File

@ -56,6 +56,7 @@ import org.apache.solr.client.solrj.impl.HttpClientUtil;
import org.apache.solr.client.solrj.impl.HttpSolrServer; import org.apache.solr.client.solrj.impl.HttpSolrServer;
import org.apache.solr.client.solrj.response.QueryResponse; import org.apache.solr.client.solrj.response.QueryResponse;
import org.apache.solr.common.SolrException; import org.apache.solr.common.SolrException;
import org.apache.solr.common.cloud.ClusterState;
import org.apache.solr.common.cloud.Replica; import org.apache.solr.common.cloud.Replica;
import org.apache.solr.common.cloud.Slice; import org.apache.solr.common.cloud.Slice;
import org.apache.solr.common.cloud.ZkCoreNodeProps; import org.apache.solr.common.cloud.ZkCoreNodeProps;
@ -869,7 +870,10 @@ public class SolrCLI {
log.info("Running healthcheck for "+collection); log.info("Running healthcheck for "+collection);
ZkStateReader zkStateReader = cloudSolrServer.getZkStateReader(); ZkStateReader zkStateReader = cloudSolrServer.getZkStateReader();
Collection<Slice> slices = zkStateReader.getClusterState().getSlices(collection);
ClusterState clusterState = zkStateReader.getClusterState();
Set<String> liveNodes = clusterState.getLiveNodes();
Collection<Slice> slices = clusterState.getSlices(collection);
if (slices == null) if (slices == null)
throw new IllegalArgumentException("Collection "+collection+" not found!"); throw new IllegalArgumentException("Collection "+collection+" not found!");
@ -908,39 +912,45 @@ public class SolrCLI {
ZkCoreNodeProps replicaCoreProps = new ZkCoreNodeProps(r); ZkCoreNodeProps replicaCoreProps = new ZkCoreNodeProps(r);
String coreUrl = replicaCoreProps.getCoreUrl(); String coreUrl = replicaCoreProps.getCoreUrl();
boolean isLeader = coreUrl.equals(leaderUrl); boolean isLeader = coreUrl.equals(leaderUrl);
// query this replica directly to get doc count and assess health
HttpSolrServer solr = new HttpSolrServer(coreUrl);
String solrUrl = solr.getBaseURL();
q = new SolrQuery("*:*");
q.setRows(0);
q.set("distrib", "false");
try {
qr = solr.query(q);
numDocs = qr.getResults().getNumFound();
int lastSlash = solrUrl.lastIndexOf('/');
String systemInfoUrl = solrUrl.substring(0,lastSlash)+"/admin/info/system";
Map<String,Object> info = getJson(solr.getHttpClient(), systemInfoUrl, 2);
uptime = uptime(asLong("/jvm/jmx/upTimeMS", info));
String usedMemory = asString("/jvm/memory/used", info);
String totalMemory = asString("/jvm/memory/total", info);
memory = usedMemory+" of "+totalMemory;
// if we get here, we can trust the state
replicaStatus = replicaCoreProps.getState();
} catch (Exception exc) {
log.error("ERROR: " + exc + " when trying to reach: " + solrUrl);
if (checkCommunicationError(exc)) { // if replica's node is not live, it's status is DOWN
replicaStatus = "down"; String nodeName = replicaCoreProps.getNodeName();
} else { if (nodeName == null || !liveNodes.contains(nodeName)) {
replicaStatus = "error: "+exc; replicaStatus = ZkStateReader.DOWN;
} } else {
} finally { // query this replica directly to get doc count and assess health
solr.shutdown(); HttpSolrServer solr = new HttpSolrServer(coreUrl);
String solrUrl = solr.getBaseURL();
q = new SolrQuery("*:*");
q.setRows(0);
q.set("distrib", "false");
try {
qr = solr.query(q);
numDocs = qr.getResults().getNumFound();
int lastSlash = solrUrl.lastIndexOf('/');
String systemInfoUrl = solrUrl.substring(0,lastSlash)+"/admin/info/system";
Map<String,Object> info = getJson(solr.getHttpClient(), systemInfoUrl, 2);
uptime = uptime(asLong("/jvm/jmx/upTimeMS", info));
String usedMemory = asString("/jvm/memory/used", info);
String totalMemory = asString("/jvm/memory/total", info);
memory = usedMemory+" of "+totalMemory;
// if we get here, we can trust the state
replicaStatus = replicaCoreProps.getState();
} catch (Exception exc) {
log.error("ERROR: " + exc + " when trying to reach: " + solrUrl);
if (checkCommunicationError(exc)) {
replicaStatus = "down";
} else {
replicaStatus = "error: "+exc;
}
} finally {
solr.shutdown();
}
} }
replicaList.add(new ReplicaHealth(shardName, r.getName(), coreUrl, replicaList.add(new ReplicaHealth(shardName, r.getName(), coreUrl,
replicaStatus, numDocs, isLeader, uptime, memory)); replicaStatus, numDocs, isLeader, uptime, memory));
} }