yexinru/doc-exports
1
0
Fork 0
forked from docs/doc-exports
Code Pull Requests Activity
Files
93d856d5c56dc6a5b27380fe001d75f16fabcfb6
doc-exports/docs/css/umn/css_01_0245.html
zhengxiu 93d856d5c5 css umn 25.6.0 version 
Reviewed-by: Pruthi, Vineet <vineet.pruthi@t-systems.com>
Co-authored-by: zhengxiu <zhengxiu@huawei.com>
Co-committed-by: zhengxiu <zhengxiu@huawei.com>
2025-11-25 11:34:43 +00:00

143 lines
30 KiB
HTML
Raw Blame History

<a name="EN-US_TOPIC_0000002038631273"></a><a name="EN-US_TOPIC_0000002038631273"></a>
<h1 class="topictitle1">Changing the Node Specifications and Storage Type of an Elasticsearch Cluster</h1>
<div id="body0000002038631273"><p id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285669680_p18220112044212">If the workloads on an Elasticsearch cluster's data plane change, you can scale the cluster vertically by changing its node specifications or node storage type.</p>
<div class="tablenoborder"><table cellpadding="4" cellspacing="0" summary="" id="EN-US_TOPIC_0000002038631273__table1447415971912" frame="border" border="1" rules="all"><caption><b>Table 1 </b>Change scenarios</caption><thead align="left"><tr id="EN-US_TOPIC_0000002038631273__row154741959161915"><th align="left" class="cellrowborder" valign="top" width="19.04809519048095%" id="mcps1.3.2.2.4.1.1"><p id="EN-US_TOPIC_0000002038631273__p84741959181916">Change Type</p>
</th>
<th align="left" class="cellrowborder" valign="top" width="40.475952404759525%" id="mcps1.3.2.2.4.1.2"><p id="EN-US_TOPIC_0000002038631273__p7474759181915">Scenario</p>
</th>
<th align="left" class="cellrowborder" valign="top" width="40.475952404759525%" id="mcps1.3.2.2.4.1.3"><p id="EN-US_TOPIC_0000002038631273__p18943191814224">Change Process</p>
</th>
</tr>
</thead>
<tbody><tr id="EN-US_TOPIC_0000002038631273__row17888195914235"><td class="cellrowborder" valign="top" width="19.04809519048095%" headers="mcps1.3.2.2.4.1.1 "><p id="EN-US_TOPIC_0000002038631273__p4420103813303">Changing node specifications</p>
</td>
<td class="cellrowborder" valign="top" width="40.475952404759525%" headers="mcps1.3.2.2.4.1.2 "><p id="EN-US_TOPIC_0000002038631273__p1485011110337">Typically, you increase node specifications instead of decreasing them. Common scenarios include:</p>
<ul id="EN-US_TOPIC_0000002038631273__ul88501013337"><li id="EN-US_TOPIC_0000002038631273__li7850511334">If the allocation of new indexes or shards takes too long or the coordination and scheduling of nodes are inefficient, increase the master node specifications.</li><li id="EN-US_TOPIC_0000002038631273__li2850218334">If too many requests need to be handled or too many results need to be aggregated, increase the client node specifications.</li><li id="EN-US_TOPIC_0000002038631273__li98501018331">If data nodes are becoming slower in responding to data writing and query requests, increase the specifications of the data nodes.</li><li id="EN-US_TOPIC_0000002038631273__li1085012103317">If cold data query becomes slow, increase the cold data node specifications.</li><li id="EN-US_TOPIC_0000002038631273__li191371522103212">If the CPU or JVM metrics of cluster nodes show performance bottlenecks, increase node specifications to improve cluster performance.</li></ul>
<p id="EN-US_TOPIC_0000002038631273__p13903191917336">Alternatively, you may also decrease node specifications, but doing so will decrease the cluster's data processing and storage capacities. Exercise caution.</p>
</td>
<td class="cellrowborder" valign="top" width="40.475952404759525%" headers="mcps1.3.2.2.4.1.3 "><ol id="EN-US_TOPIC_0000002038631273__ol647416302517"><li id="EN-US_TOPIC_0000002038631273__li84741430165117">Bring a node offline.</li><li id="EN-US_TOPIC_0000002038631273__li1147423016518">Change the node specifications.</li><li id="EN-US_TOPIC_0000002038631273__li20333641115915">Restart the node and restore its data.</li><li id="EN-US_TOPIC_0000002038631273__li14474330135117">After the node recovers, proceed to another node and repeat the steps above. This goes on until all nodes are changed.</li></ol>
<p id="EN-US_TOPIC_0000002038631273__p826110469513">The node specifications are changed one node at a time. This is to ensure that there are sufficient resources to keep services running.</p>
</td>
</tr>
<tr id="EN-US_TOPIC_0000002038631273__row105781718244"><td class="cellrowborder" valign="top" width="19.04809519048095%" headers="mcps1.3.2.2.4.1.1 "><p id="EN-US_TOPIC_0000002038631273__p10416203863013">Changing the node storage type (disk type)</p>
</td>
<td class="cellrowborder" valign="top" width="40.475952404759525%" headers="mcps1.3.2.2.4.1.2 "><p id="EN-US_TOPIC_0000002038631273__p11950101817713">Change the node storage type if disk I/O has become a performance bottleneck, which impacts query and write performance.</p>
</td>
<td class="cellrowborder" valign="top" width="40.475952404759525%" headers="mcps1.3.2.2.4.1.3 "><ol id="EN-US_TOPIC_0000002038631273__ol152470495382"><li id="EN-US_TOPIC_0000002038631273__li1026844812446">Select a random node and migrate its data to other nodes.</li><li id="EN-US_TOPIC_0000002038631273__li626894814446">Rebuild the node using the new storage type.</li><li id="EN-US_TOPIC_0000002038631273__li712112419112">Add the node back into the cluster. The system automatically triggers a shard reallocation, migrating some of the shards to the new node.</li><li id="EN-US_TOPIC_0000002038631273__li126812484443">After the node recovers, proceed to another node and repeat the steps above.</li></ol>
<p id="EN-US_TOPIC_0000002038631273__p2705131010386">The nodes are changed one at a time to prevent service interruptions.</p>
<p id="EN-US_TOPIC_0000002038631273__p157010795312"></p>
</td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section19365446164319"><h4 class="sectiontitle">Impact on Billing</h4><p id="EN-US_TOPIC_0000002038631273__p1525719335610">For a pay-per-use cluster, you can see its new price when confirming the node specifications or storage type change on the console. After the change is complete, the new price will apply. For pricing details, see .</p>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section1097175124210"><h4 class="sectiontitle">Constraints</h4><ul id="EN-US_TOPIC_0000002038631273__ul66998497299"><li id="EN-US_TOPIC_0000002038631273__li469934912912">The node specifications and storage type cannot be changed for nodes that use local disks.</li><li id="EN-US_TOPIC_0000002038631273__li16699144911297">The node specifications and storage type cannot be changed at the same time.</li><li id="EN-US_TOPIC_0000002038631273__li156993498297">The node storage type can be changed only for data nodes and cold data nodes.</li><li id="EN-US_TOPIC_0000002038631273__li1929617202416">When you change the node storage type, data needs to be migrated between different nodes. The timeout for data migration per node is 48 hours. Upgrade will fail if this timeout expires. When the cluster has large quantities of data, you are advised to manually adjust the data migration rate and avoid performing the migration during peak hours.</li><li id="EN-US_TOPIC_0000002038631273__li6699154982916">For a cluster without master nodes, the node storage type can be changed only if the number of data nodes plus cold data nodes is at least three.</li><li id="EN-US_TOPIC_0000002038631273__li1069919494292">For a cluster with master nodes, this operation is allowed only if the cluster has at least two data nodes.</li><li id="EN-US_TOPIC_0000002038631273__li13699184932912">During a node storage type change, there is always one node that is unavailable. To ensure service continuity, make sure the <strong id="EN-US_TOPIC_0000002038631273__b974064119215">total number of data nodes and cold data nodes is greater than the maximum number of index replicas plus 1</strong>. For a single-AZ or dual-AZ cluster, also make sure there are at least two nodes of each type for the cluster in each of its AZs.</li><li id="EN-US_TOPIC_0000002038631273__li76991849162910">During a node specifications change, nodes are brought offline in order to make the changes. To ensure service continuity, make sure all shards have replicas.</li><li id="EN-US_TOPIC_0000002038631273__li12554142393213">Make sure the disk usage is always less than 80% during the change.</li></ul>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section12392055185715"><h4 class="sectiontitle">Change Impact</h4><p id="EN-US_TOPIC_0000002038631273__p32333115589">Before changing a cluster's node specifications or storage type, it is essential to assess the potential impacts and review operational recommendations. This enables proper scheduling of the change, minimizing service interruptions.</p>
<ul id="EN-US_TOPIC_0000002038631273__ul10438182715118"><li id="EN-US_TOPIC_0000002038631273__li1999951063217">Performance impact (only when the node storage type is changed)<p id="EN-US_TOPIC_0000002038631273__p12755222133911"><a name="EN-US_TOPIC_0000002038631273__li1999951063217"></a><a name="li1999951063217"></a>Changing the node storage type does not interrupt services. However, data migration that occurs during this process consumes I/O performance, and taking individual nodes offline still has some impact on the overall cluster performance.</p>
<div class="p" id="EN-US_TOPIC_0000002038631273__p16986568357">To minimize this impact, it is advisable to adjust the data migration rate based on the cluster's traffic cycle: increase the data migration rate during off-peak hours to shorten the task duration, and decrease it <strong id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001965496905_b7158153214314">before</strong> peak hours arrive to ensure optimal cluster performance. The data migration rate is determined by the <span class="parmname" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001965496905_parmname525325404618"><b>indices.recovery.max_bytes_per_sec</b></span> parameter. The default value of this parameter is the number of vCPUs multiplied by 32 MB. For example, for four vCPUs, the data migration rate is 128 MB. Set this parameter to a value between 40 MB and 1000 MB based on your service requirements.<pre class="screen" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001965496905_screen67611747479">PUT /_cluster/settings
{
"transient": {
"indices.recovery.max_bytes_per_sec": "1000MB"
}
}</pre>
</div>
</li><li id="EN-US_TOPIC_0000002038631273__li643812278112">Impact on request handling<p id="EN-US_TOPIC_0000002038631273__p163869181212"><a name="EN-US_TOPIC_0000002038631273__li643812278112"></a><a name="li643812278112"></a>Taking nodes offline one at a time usually does not interrupt services. However, requests sent to offline nodes may fail. To mitigate this impact, the following measures may be taken:</p>
<ul id="EN-US_TOPIC_0000002038631273__ul1991315132140"><li id="EN-US_TOPIC_0000002038631273__li4662191251411">Use a VPC endpoint or a dedicated load balancer to handle access requests to your cluster, which makes sure that requests are automatically routed to available nodes.</li><li id="EN-US_TOPIC_0000002038631273__li56621012111410">Enable an exponential backoff &amp; retry mechanism on the client (configure three retries).</li><li id="EN-US_TOPIC_0000002038631273__li1866320129142">Perform this operation during off-peak hours.</li></ul>
</li><li id="EN-US_TOPIC_0000002038631273__li124091516101417">Impact on index replicas<p id="EN-US_TOPIC_0000002038631273__p16415141681418"><a name="EN-US_TOPIC_0000002038631273__li124091516101417"></a><a name="li124091516101417"></a>Shards that have no replicas will become unavailable when the nodes that store them are taken offline, causing service interruptions. You are advised to add replicas for all important indexes before making the change described in this topic.</p>
</li><li id="EN-US_TOPIC_0000002038631273__li10787555122719">Impact on Kibana and Cerebro<p id="EN-US_TOPIC_0000002038631273__p127871655102717"><a name="EN-US_TOPIC_0000002038631273__li10787555122719"></a><a name="li10787555122719"></a>Changing the node storage type for a cluster will cause Kibana and Cerebro to be rebuilt. During this period, Kibana and Cerebro are temporarily unavailable. During a node specifications change, if Kibana and Cerebro become unavailable because the node that runs them is taken offline, refresh the web page or try to log in again, and the system will reschedule Kibana and Cerebro to an available node.</p>
</li><li id="EN-US_TOPIC_0000002038631273__li1697691517153">Characteristics of this process<p id="EN-US_TOPIC_0000002038631273__p1297941510158"><a name="EN-US_TOPIC_0000002038631273__li1697691517153"></a><a name="li1697691517153"></a>Once started, a change task cannot be stopped until it succeeds or fails. A change task failure only impacts a single node, and does not interrupt services if there are data replicas, but the failed node still needs to be restored promptly.</p>
</li></ul>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section1147814313341"><h4 class="sectiontitle">Change Durations</h4><ul id="EN-US_TOPIC_0000002038631273__ul259914593719"><li id="EN-US_TOPIC_0000002038631273__li55991759275">The following formulas can be used to estimate how long the node specifications change will take:<p id="EN-US_TOPIC_0000002038631273__p2382153812219"><a name="EN-US_TOPIC_0000002038631273__li55991759275"></a><a name="li55991759275"></a><strong id="EN-US_TOPIC_0000002038631273__b7509195020138">Change duration (min) = 10 (min) x Total number of nodes to change + Data recovery duration (min)</strong></p>
<div class="p" id="EN-US_TOPIC_0000002038631273__p13772163437">where,<ul id="EN-US_TOPIC_0000002038631273__ul187731238317"><li id="EN-US_TOPIC_0000002038631273__li1477319319320">10 minutes indicates how long non-data recovery operations (e.g., initialization) typically take per node. It is an empirical value.</li><li id="EN-US_TOPIC_0000002038631273__li157731432318">The total number of nodes is the sum of the number of data nodes, master nodes, client nodes, and cold data nodes in the cluster.</li></ul>
</div>
<p id="EN-US_TOPIC_0000002038631273__p1777595312315"><strong id="EN-US_TOPIC_0000002038631273__b24711960154">Data recovery duration (min) = Total data size (MB)/[Total number of vCPUs of the data nodes x 32 (MB/s) x 60 (s)]</strong></p>
<div class="p" id="EN-US_TOPIC_0000002038631273__p28141516068">where,<ul id="EN-US_TOPIC_0000002038631273__ul115291993613"><li id="EN-US_TOPIC_0000002038631273__li1452919766">32 MB/s indicates that each vCPU can process 32 MB of data per second. It is an empirical value.</li><li id="EN-US_TOPIC_0000002038631273__li14529891964">The formulas above use estimates under ideal conditions. The actual data recovery speed depends on cluster load.</li></ul>
</div>
</li></ul>
<ul id="EN-US_TOPIC_0000002038631273__ul1685716518814"><li id="EN-US_TOPIC_0000002038631273__li198571855815">The following formulas can be used to estimate how long the node storage type change will take:<p id="EN-US_TOPIC_0000002038631273__p385775383"><a name="EN-US_TOPIC_0000002038631273__li198571855815"></a><a name="li198571855815"></a><strong id="EN-US_TOPIC_0000002038631273__b19086044653118">Change duration (min) = 15 (min) x Total number of nodes to change + Data migration duration (min)</strong></p>
<div class="p" id="EN-US_TOPIC_0000002038631273__p15857155819">where,<ul id="EN-US_TOPIC_0000002038631273__ul8857155188"><li id="EN-US_TOPIC_0000002038631273__li16857165887">15 minutes indicates how long non-data migration operations (e.g., initialization) typically take per node. It is an empirical value.</li><li id="EN-US_TOPIC_0000002038631273__li1285715516819">The total number of nodes is the sum of the number of data nodes, master nodes, client nodes, and cold data nodes in the cluster.</li></ul>
</div>
<p id="EN-US_TOPIC_0000002038631273__p78571516815"><strong id="EN-US_TOPIC_0000002038631273__b221012819271">Data migration duration (min) = Total data size (MB)/[Total number of vCPUs of the data nodes x 32 (MB/s) x 60 (s)]</strong></p>
<div class="p" id="EN-US_TOPIC_0000002038631273__p158581251884">where,<ul id="EN-US_TOPIC_0000002038631273__ul178589516811"><li id="EN-US_TOPIC_0000002038631273__li2858153817">32 MB/s indicates that each vCPU can process 32 MB of data per second. It is an empirical value.</li><li id="EN-US_TOPIC_0000002038631273__li2858353818">The formulas above use estimates under ideal conditions. The actual migration speed depends on cluster load.</li></ul>
</div>
</li></ul>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section71501434164515"><h4 class="sectiontitle">Prerequisites</h4><ul id="EN-US_TOPIC_0000002038631273__ul5150183414451"><li id="EN-US_TOPIC_0000002038631273__li1115093415451">The cluster status is <strong id="EN-US_TOPIC_0000002038631273__b13747132112415">Available</strong>, and there are no ongoing tasks.</li><li id="EN-US_TOPIC_0000002038631273__li1815011348456">Your CSS resource quotas are sufficient for the capacity expansion you are about to perform. You can check available resources on the <strong id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001938218180_b1073163610569">Modify Configuration</strong> page.</li><li id="EN-US_TOPIC_0000002038631273__li1715111347458">All mission-critical data has been backed up before a node storage type change. This is to prevent data loss. For details, see <a href="css_01_0267.html">Creating Snapshots to Back Up the Data of an Elasticsearch Cluster</a>.</li></ul>
</div>
<div class="section" id="EN-US_TOPIC_0000002038631273__section186091514113516"><h4 class="sectiontitle">Changing the Node Specifications and Storage Type</h4><ol id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_ol373518361145"><li id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_li1173513611411"><span id="EN-US_TOPIC_0000002038631273__ph4222205142920">Log in to the CSS management console.</span></li><li id="EN-US_TOPIC_0000002038631273__li4250043448">In the navigation pane on the left, choose <span class="uicontrol" id="EN-US_TOPIC_0000002038631273__uicontrol14290155318141"><b>Clusters &gt; Elasticsearch</b></span>.</li><li id="EN-US_TOPIC_0000002038631273__li163611336525">Check that all service data has replicas so that services will not be interrupted during the change.<ol type="a" id="EN-US_TOPIC_0000002038631273__ol202982040105515"><li id="EN-US_TOPIC_0000002038631273__li1404113845517">In the displayed cluster list, find the target cluster, and click <strong id="EN-US_TOPIC_0000002038631273__b1978191411717">Access Kibana</strong> in the <strong id="EN-US_TOPIC_0000002038631273__b1378114146176">Operation</strong> column to log in to the Kibana console.</li><li id="EN-US_TOPIC_0000002038631273__li1034524695516">In the left navigation pane, choose <strong id="EN-US_TOPIC_0000002038631273__b9885021191712">Dev Tools</strong>.</li><li id="EN-US_TOPIC_0000002038631273__li156519532559">Run the <b><span class="cmdname" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_cmdname14352184252419">GET _cat/indices?v</span></b> command in Kibana.<ul id="EN-US_TOPIC_0000002038631273__ul28380564557"><li id="EN-US_TOPIC_0000002038631273__li1388213095616">If the returned <strong id="EN-US_TOPIC_0000002038631273__b111271195581218">rep</strong> value is greater than <strong id="EN-US_TOPIC_0000002038631273__b97037799781218">0</strong>, data replicas exist. Go to the next step.</li><li id="EN-US_TOPIC_0000002038631273__li18381056175520">If the returned <strong id="EN-US_TOPIC_0000002038631273__b12948397154">rep</strong> value is 0, there are no data replicas. You are advised to manually create a snapshot for the cluster before moving on to the next step. For details, see <a href="css_01_0267.html">Creating Snapshots to Back Up the Data of an Elasticsearch Cluster</a>.</li></ul>
</li></ol>
</li><li id="EN-US_TOPIC_0000002038631273__li01685973618">In the cluster list, find the target cluster, and choose <strong id="EN-US_TOPIC_0000002038631273__b103718643591921">More</strong> &gt; <strong id="EN-US_TOPIC_0000002038631273__b204392001391921">Modify Configuration</strong> in the <strong id="EN-US_TOPIC_0000002038631273__b116592812791921">Operation</strong> column. The <strong id="EN-US_TOPIC_0000002038631273__b23523264191921">Modify Configuration</strong> page is displayed.</li><li id="EN-US_TOPIC_0000002038631273__li1757812586252">Click the <strong id="EN-US_TOPIC_0000002038631273__b122097225981218">Scale Cluster</strong> tab.</li><li id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_li13735736141416">Click <strong id="EN-US_TOPIC_0000002038631273__b198787908381218">Change specifications</strong> to set parameters.
<div class="tablenoborder"><table cellpadding="4" cellspacing="0" summary="" id="EN-US_TOPIC_0000002038631273__table5252102312527" frame="border" border="1" rules="all"><caption><b>Table 2 </b>Changing the node specifications and storage type</caption><thead align="left"><tr id="EN-US_TOPIC_0000002038631273__row1125202316521"><th align="left" class="cellrowborder" valign="top" width="30%" id="mcps1.3.8.2.6.2.2.3.1.1"><p id="EN-US_TOPIC_0000002038631273__p15252623105212">Parameter</p>
</th>
<th align="left" class="cellrowborder" valign="top" width="70%" id="mcps1.3.8.2.6.2.2.3.1.2"><p id="EN-US_TOPIC_0000002038631273__p7252142305220">Description</p>
</th>
</tr>
</thead>
<tbody><tr id="EN-US_TOPIC_0000002038631273__row1025322310525"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.6.2.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p4253122375216">Action</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.6.2.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p1925372314529">Select <strong id="EN-US_TOPIC_0000002038631273__b170544963581218">Change specifications</strong>.</p>
</td>
</tr>
<tr id="EN-US_TOPIC_0000002038631273__row8253152318527"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.6.2.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p1125318232526">Resources</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.6.2.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p14253423125216">Shows the change of resources for this operation.</p>
</td>
</tr>
<tr id="EN-US_TOPIC_0000002038631273__row1425318234522"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.6.2.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p62535237520">Nodes</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.6.2.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p669218165118">Configure the changes you want to make.</p>
<ol type="a" id="EN-US_TOPIC_0000002038631273__ol319592411216"><li id="EN-US_TOPIC_0000002038631273__li16195142414129">Select a node type in the <span class="parmvalue" id="EN-US_TOPIC_0000002038631273__parmvalue11570300783118"><b>Node Type</b></span> column.</li><li id="EN-US_TOPIC_0000002038631273__li219514248126">Select a new flavor in the <strong id="EN-US_TOPIC_0000002038631273__b1422934720219">Node Specifications</strong> column, or select a new storage type in the <strong id="EN-US_TOPIC_0000002038631273__b123017474210">Node Storage</strong> column.</li></ol>
<p id="EN-US_TOPIC_0000002038631273__p473431441110">The node specifications and storage type cannot be changed at the same time.</p>
</td>
</tr>
</tbody>
</table>
</div>
</li><li id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_li173623641418">Click <span class="uicontrol" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_uicontrol073673619146"><b>Next</b></span>.</li><li id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_li47662104588">Confirm the information and click <strong id="EN-US_TOPIC_0000002038631273__b45645173681218">Submit</strong>.</li><li id="EN-US_TOPIC_0000002038631273__li533453111613">In the displayed dialog box, confirm check items, and click <strong id="EN-US_TOPIC_0000002038631273__b16911556204018">OK</strong> to start the specifications change.<ul id="EN-US_TOPIC_0000002038631273__ul847915412168"><li id="EN-US_TOPIC_0000002038631273__li547915419167">Check items for node specifications change: <strong id="EN-US_TOPIC_0000002038631273__b179001989429">Verify index copies</strong> and <strong id="EN-US_TOPIC_0000002038631273__b1517911204217">Cluster status check</strong>.</li><li id="EN-US_TOPIC_0000002038631273__li4938514131712">Check items for node storage type change: <strong id="EN-US_TOPIC_0000002038631273__b178881625134216">Check cluster load</strong>.</li></ul>
<div class="tablenoborder"><table cellpadding="4" cellspacing="0" summary="" id="EN-US_TOPIC_0000002038631273__table697119186181" frame="border" border="1" rules="all"><caption><b>Table 3 </b>Check item description</caption><thead align="left"><tr id="EN-US_TOPIC_0000002038631273__row2971218121816"><th align="left" class="cellrowborder" valign="top" width="30%" id="mcps1.3.8.2.9.3.2.3.1.1"><p id="EN-US_TOPIC_0000002038631273__p6971121841814">Item</p>
</th>
<th align="left" class="cellrowborder" valign="top" width="70%" id="mcps1.3.8.2.9.3.2.3.1.2"><p id="EN-US_TOPIC_0000002038631273__p11971181861811">Description</p>
</th>
</tr>
</thead>
<tbody><tr id="EN-US_TOPIC_0000002038631273__row697121801813"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.9.3.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p6971121818183">Verify index copies</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.9.3.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p1971118131815">By default, CSS checks for indexes that do not have any replicas created for them. You can skip this step, but the lack of index replicas may impact service availability during a node specifications change.</p>
<ul id="EN-US_TOPIC_0000002038631273__ul16560184362314"><li id="EN-US_TOPIC_0000002038631273__li956024312316">If you select <strong id="EN-US_TOPIC_0000002038631273__b43760772381218">Verify index copies</strong> and the cluster has no master nodes, each index is required to have at least one replica and the cluster must have at least three nodes.</li><li id="EN-US_TOPIC_0000002038631273__li3560134352311">If you select <strong id="EN-US_TOPIC_0000002038631273__b178807832481218">Verify index copies</strong> and the cluster has a master node, each index is required to have at least one replica, but there is no requirement on how many nodes the cluster must have.</li></ul>
</td>
</tr>
<tr id="EN-US_TOPIC_0000002038631273__row149713181185"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.9.3.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p19971518201811">Cluster status check</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.9.3.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p355813407611">During a node specifications change, the cluster status is checked by default to improve the success rate and ensure data security. The nodes are changed one at a time. For each node, the system changes its specifications, restarts it, and checks that all its processes are started successfully before moving on to the next node.</p>
<p id="EN-US_TOPIC_0000002038631273__p1455918401861">In emergencies (for example, when a cluster is overloaded and services are faulty, which may prevent a specifications change request from being delivered), you can skip cluster status check so that more resources can be made available for cluster recovery. However, doing so may cause the cluster to become faulty and interrupt services. Exercise caution.</p>
</td>
</tr>
<tr id="EN-US_TOPIC_0000002038631273__row589204210182"><td class="cellrowborder" valign="top" width="30%" headers="mcps1.3.8.2.9.3.2.3.1.1 "><p id="EN-US_TOPIC_0000002038631273__p108974213188">Check cluster load</p>
</td>
<td class="cellrowborder" valign="top" width="70%" headers="mcps1.3.8.2.9.3.2.3.1.2 "><p id="EN-US_TOPIC_0000002038631273__p389942141815">During a node storage type change, data migration between nodes and the stopping and restarting of nodes will consume cluster resources, causing the cluster load to increase. A cluster load check can identify possible overload risks for a cluster and reduce the likelihood of an overload condition causing the node storage type change to fail.</p>
<p id="EN-US_TOPIC_0000002038631273__p13159136527">The cluster load check items are as follows:</p>
<ul id="EN-US_TOPIC_0000002038631273__ul103491858155210"><li id="EN-US_TOPIC_0000002038631273__li4349258135212"><strong id="EN-US_TOPIC_0000002038631273__b68451335193812">nodes.thread_pool.search.queue &lt; 1000</strong>: Check that the maximum number of requests in the search queue is less than 1000.</li><li id="EN-US_TOPIC_0000002038631273__li0349358195215"><strong id="EN-US_TOPIC_0000002038631273__b2096745518386">nodes.thread_pool.write.queue &lt; 200</strong>: Check that the maximum number of requests in the write queue is less than 200.</li><li id="EN-US_TOPIC_0000002038631273__li103491058105217"><strong id="EN-US_TOPIC_0000002038631273__b285715127394">nodes.process.cpu.percent &lt; 90</strong>: Check that the maximum CPU usage is less than 90%.</li><li id="EN-US_TOPIC_0000002038631273__li2349105805215"><strong id="EN-US_TOPIC_0000002038631273__b1696542316395">nodes.os.cpu.load_average/Number of vCPUs &lt; 80%</strong>: Check that the number of running processes plus the number of processes waiting for CPUs is less than 80% of the total number of vCPUs.</li></ul>
</td>
</tr>
</tbody>
</table>
</div>
<div class="note" id="EN-US_TOPIC_0000002038631273__note1882644712146"><img src="public_sys-resources/note_3.0-en-us.png"><span class="notetitle"> </span><div class="notebody"><p id="EN-US_TOPIC_0000002038631273__p188261477148">If the change request fails to be submitted and a message is displayed indicating that the cluster needs to be upgraded, it means the current cluster version does not support a node storage type change. Upgrade the cluster to the latest image version and then try again. For a detailed upgrade guide, see <a href="css_01_0186.html">Upgrading the Version of an Elasticsearch Cluster</a>.</p>
</div></div>
</li><li id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_li13737436141418">Click <span class="uicontrol" id="EN-US_TOPIC_0000002038631273__uicontrol188375124691848"><b>Back to Cluster List</b></span> to go back to the <strong id="EN-US_TOPIC_0000002038631273__b23654524991848">Clusters</strong> page. The <span class="parmname" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_parmname1274216574249"><b>Cluster Status</b></span> is <span class="parmvalue" id="EN-US_TOPIC_0000002038631273__en-us_topic_0000001285509756_parmvalue18742135702419"><b>Configuration modified</b></span>. When <span class="parmname" id="EN-US_TOPIC_0000002038631273__parmname16215673081218"><b>Cluster Status</b></span> changes to <span class="parmvalue" id="EN-US_TOPIC_0000002038631273__parmvalue96562252381218"><b>Available</b></span>, the cluster node specifications have been successfully changed.</li></ol>
</div>
</div>
<div>
<div class="familylinks">
<div class="parentlink"><strong>Parent topic:</strong> <a href="css_01_0009.html">Managing Elasticsearch Clusters</a></div>
</div>
</div>
View Git Blame Copy Permalink