📄 crawlcontroller.html
字号:
<a name="418" href="#418">418</a> throws IOException {<a name="419" href="#419">419</a> <strong>long</strong> started = System.currentTimeMillis();;<a name="420" href="#420">420</a> <strong>if</strong> (LOGGER.isLoggable(Level.FINE)) {<a name="421" href="#421">421</a> LOGGER.fine(<span class="string">"Starting recovery setup -- copying into place "</span> +<a name="422" href="#422">422</a> <span class="string">"bdbje log files -- for checkpoint named "</span> +<a name="423" href="#423">423</a> <strong>this</strong>.checkpointRecover.getDisplayName());<a name="424" href="#424">424</a> }<a name="425" href="#425">425</a> <em class="comment">// Mark context we're in a recovery.</em><a name="426" href="#426">426</a> <strong>this</strong>.checkpointer.recover(<strong>this</strong>);<a name="427" href="#427">427</a> <strong>this</strong>.progressStats.info(<span class="string">"CHECKPOINT RECOVER "</span> +<a name="428" href="#428">428</a> <strong>this</strong>.checkpointRecover.getDisplayName());<a name="429" href="#429">429</a> <em class="comment">// Copy the bdb log files to the state dir so we don't damage</em><a name="430" href="#430">430</a> <em class="comment">// old checkpoint. If thousands of log files, can take</em><a name="431" href="#431">431</a> <em class="comment">// tens of minutes (1000 logs takes ~5 minutes to java copy,</em><a name="432" href="#432">432</a> <em class="comment">// dependent upon hardware). If log file already exists over in the</em><a name="433" href="#433">433</a> <em class="comment">// target state directory, we do not overwrite -- we assume the log</em><a name="434" href="#434">434</a> <em class="comment">// file in the target same as one we'd copy from the checkpoint dir.</em><a name="435" href="#435">435</a> File bdbSubDir = CheckpointUtils.<a name="436" href="#436">436</a> getBdbSubDirectory(<strong>this</strong>.checkpointRecover.getDirectory());<a name="437" href="#437">437</a> FileUtils.copyFiles(bdbSubDir, CheckpointUtils.getJeLogsFilter(),<a name="438" href="#438">438</a> getStateDisk(), <strong>true</strong>,<a name="439" href="#439">439</a> false);<a name="440" href="#440">440</a> <strong>if</strong> (LOGGER.isLoggable(Level.INFO)) {<a name="441" href="#441">441</a> LOGGER.info(<span class="string">"Finished recovery setup for checkpoint named "</span> +<a name="442" href="#442">442</a> <strong>this</strong>.checkpointRecover.getDisplayName() + <span class="string">" in "</span> +<a name="443" href="#443">443</a> (System.currentTimeMillis() - started) + <span class="string">"ms."</span>);<a name="444" href="#444">444</a> }<a name="445" href="#445">445</a> }<a name="446" href="#446">446</a> <a name="447" href="#447">447</a> <strong>protected</strong> <strong>boolean</strong> getCheckpointCopyBdbjeLogs() {<a name="448" href="#448">448</a> <strong>return</strong> ((Boolean)<strong>this</strong>.order.getUncheckedAttribute(<strong>null</strong>,<a name="449" href="#449">449</a> CrawlOrder.ATTR_CHECKPOINT_COPY_BDBJE_LOGS)).booleanValue();<a name="450" href="#450">450</a> }<a name="451" href="#451">451</a> <a name="452" href="#452">452</a> <strong>private</strong> <strong>void</strong> setupBdb()<a name="453" href="#453">453</a> throws FatalConfigurationException, AttributeNotFoundException {<a name="454" href="#454">454</a> EnvironmentConfig envConfig = <strong>new</strong> EnvironmentConfig();<a name="455" href="#455">455</a> envConfig.setAllowCreate(<strong>true</strong>);<a name="456" href="#456">456</a> <strong>int</strong> bdbCachePercent = ((Integer)<strong>this</strong>.order.<a name="457" href="#457">457</a> getAttribute(<strong>null</strong>, CrawlOrder.ATTR_BDB_CACHE_PERCENT)).intValue();<a name="458" href="#458">458</a> <strong>if</strong>(bdbCachePercent > 0) {<a name="459" href="#459">459</a> <em class="comment">// Operator has expressed a preference; override BDB default or </em><a name="460" href="#460">460</a> <em class="comment">// je.properties value</em><a name="461" href="#461">461</a> envConfig.setCachePercent(bdbCachePercent);<a name="462" href="#462">462</a> }<a name="463" href="#463">463</a> envConfig.setLockTimeout(5000000); <em class="comment">// 5 seconds</em><a name="464" href="#464">464</a> <strong>if</strong> (LOGGER.isLoggable(Level.FINEST)) {<a name="465" href="#465">465</a> envConfig.setConfigParam(<span class="string">"java.util.logging.level"</span>, <span class="string">"SEVERE"</span>);<a name="466" href="#466">466</a> envConfig.setConfigParam(<span class="string">"java.util.logging.level.evictor"</span>,<a name="467" href="#467">467</a> <span class="string">"SEVERE"</span>);<a name="468" href="#468">468</a> envConfig.setConfigParam(<span class="string">"java.util.logging.ConsoleHandler.on"</span>,<a name="469" href="#469">469</a> <span class="string">"true"</span>);<a name="470" href="#470">470</a> }<a name="471" href="#471">471</a> <a name="472" href="#472">472</a> <strong>if</strong> (!getCheckpointCopyBdbjeLogs()) {<a name="473" href="#473">473</a> <em class="comment">// If we are not copying files on checkpoint, then set bdbje to not</em><a name="474" href="#474">474</a> <em class="comment">// remove its log files so that its possible to later assemble</em><a name="475" href="#475">475</a> <em class="comment">// (manually) all needed to run a recovery using mix of current</em><a name="476" href="#476">476</a> <em class="comment">// bdbje logs and those its marked for deletion.</em><a name="477" href="#477">477</a> envConfig.setConfigParam(<span class="string">"je.cleaner.expunge"</span>, <span class="string">"false"</span>);<a name="478" href="#478">478</a> }<a name="479" href="#479">479</a> <a name="480" href="#480">480</a> <strong>try</strong> {<a name="481" href="#481">481</a> <strong>this</strong>.bdbEnvironment = <strong>new</strong> Environment(getStateDisk(), envConfig);<a name="482" href="#482">482</a> <strong>if</strong> (LOGGER.isLoggable(Level.FINE)) {<a name="483" href="#483">483</a> <em class="comment">// Write out the bdb configuration.</em><a name="484" href="#484">484</a> envConfig = bdbEnvironment.getConfig();<a name="485" href="#485">485</a> LOGGER.fine(<span class="string">"BdbConfiguration: Cache percentage "</span> +<a name="486" href="#486">486</a> envConfig.getCachePercent() +<a name="487" href="#487">487</a> <span class="string">", cache size "</span> + envConfig.getCacheSize());<a name="488" href="#488">488</a> }<a name="489" href="#489">489</a> <em class="comment">// Open the class catalog database. Create it if it does not</em><a name="490" href="#490">490</a> <em class="comment">// already exist. </em><a name="491" href="#491">491</a> DatabaseConfig dbConfig = <strong>new</strong> DatabaseConfig();<a name="492" href="#492">492</a> dbConfig.setAllowCreate(<strong>true</strong>);<a name="493" href="#493">493</a> <strong>this</strong>.classCatalogDB = <strong>this</strong>.bdbEnvironment.<a name="494" href="#494">494</a> openDatabase(<strong>null</strong>, <span class="string">"classes"</span>, dbConfig);<a name="495" href="#495">495</a> <strong>this</strong>.classCatalog = <strong>new</strong> StoredClassCatalog(classCatalogDB);<a name="496" href="#496">496</a> } <strong>catch</strong> (DatabaseException e) {<a name="497" href="#497">497</a> e.printStackTrace();<a name="498" href="#498">498</a> <strong>throw</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/framework/exceptions/FatalConfigurationException.html">FatalConfigurationException</a>(e.getMessage());<a name="499" href="#499">499</a> }<a name="500" href="#500">500</a> }<a name="501" href="#501">501</a> <a name="502" href="#502">502</a> <strong>public</strong> Environment getBdbEnvironment() {<a name="503" href="#503">503</a> <strong>return</strong> <strong>this</strong>.bdbEnvironment;<a name="504" href="#504">504</a> }<a name="505" href="#505">505</a> <a name="506" href="#506">506</a> <strong>public</strong> StoredClassCatalog getClassCatalog() {<a name="507" href="#507">507</a> <strong>return</strong> <strong>this</strong>.classCatalog;<a name="508" href="#508">508</a> }<a name="509" href="#509">509</a> <a name="510" href="#510">510</a> <em>/**<em>*</em></em><a name="511" href="#511">511</a> <em> * Register for CrawlStatus events.</em><a name="512" href="#512">512</a> <em> *</em><a name="513" href="#513">513</a> <em> * @param cl a class implementing the CrawlStatusListener interface</em><a name="514" href="#514">514</a> <em> *</em><a name="515" href="#515">515</a> <em> * @see CrawlStatusListener</em><a name="516" href="#516">516</a> <em> */</em><a name="517" href="#517">517</a> <strong>public</strong> <strong>void</strong> addCrawlStatusListener(<a href="../../../../org/archive/crawler/event/CrawlStatusListener.html">CrawlStatusListener</a> cl) {<a name="518" href="#518">518</a> <strong>synchronized</strong> (<strong>this</strong>.registeredCrawlStatusListeners) {<a name="519" href="#519">519</a> <strong>this</strong>.registeredCrawlStatusListeners.add(cl);<a name="520" href="#520">520</a> }<a name="521" href="#521">521</a> }<a name="522" href="#522">522</a> <a name="523" href="#523">523</a> <em>/**<em>*</em></em><a name="524" href="#524">524</a> <em> * Register for CrawlURIDisposition events.</em><a name="525" href="#525">525</a> <em> *</em><a name="526" href="#526">526</a> <em> * @param cl a class implementing the CrawlURIDispostionListener interface</em><a name="527" href="#527">527</a> <em> *</em><a name="528" href="#528">528</a> <em> * @see CrawlURIDispositionListener</em><a name="529" href="#529">529</a> <em> */</em><a name="530" href="#530">530</a> <strong>public</strong> <strong>void</strong> addCrawlURIDispositionListener(<a href="../../../../org/archive/crawler/event/CrawlURIDispositionListener.html">CrawlURIDispositionListener</a> cl) {<a name="531" href="#531">531</a> registeredCrawlURIDispositionListener = <strong>null</strong>;<a name="532" href="#532">532</a> <strong>if</strong> (registeredCrawlURIDispositionListeners == <strong>null</strong>) {<a name="533" href="#533">533</a> <em class="comment">// First listener;</em><a name="534" href="#534">534</a> registeredCrawlURIDispositionListener = cl;<a name="535" href="#535">535</a> <em class="comment">//Only used for the first one while it is the only one.</em><a name="536" href="#536">536</a> registeredCrawlURIDispositionListeners = <strong>new</strong> ArrayList(1);<a name="537" href="#537">537</a> <em class="comment">//We expect it to be very small.</em><a name="538" href="#538">538</a> }<a name="539" href="#539">539</a> registeredCrawlURIDispositionListeners.add(cl);<a name="540" href="#540">540</a> }<a name="541" href="#541">541</a> <a name="542" href="#542">542</a> <em>/**<em>*</em></em><a name="543" href="#543">543</a> <em> * Allows an external class to raise a CrawlURIDispostion</em><a name="544" href="#544">544</a> <em> * crawledURISuccessful event that will be broadcast to all listeners that</em><a name="545" href="#545">545</a> <em> * have registered with the CrawlController.</em><a name="546" href="#546">546</a> <em> *</em><a name="547" href="#547">547</a> <em> * @param curi - The CrawlURI that will be sent with the event notification.</em><a name="548" href="#548">548</a> <em> *</em><a name="549" href="#549">549</a> <em> * @see CrawlURIDispositionListener#crawledURISuccessful(CrawlURI)</em><a name="550" href="#550">550</a> <em> */</em><a name="551" href="#551">551</a> <strong>public</strong> <strong>void</strong> fireCrawledURISuccessfulEvent(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="552" href="#552">552</a> <strong>if</strong> (registeredCrawlURIDispositionListener != <strong>null</strong>) {<a name="553" href="#553">553</a> <em class="comment">// Then we'll just use that.</em><a name="554" href="#554">554</a> registeredCrawlURIDispositionListener.crawledURISuccessful(curi);<a name="555" href="#555">555</a> } <strong>else</strong> {<a name="556" href="#556">556</a> <em class="comment">// Go through the list.</em><a name="557" href="#557">557</a> <strong>if</strong> (registeredCrawlURIDispositionListeners != <strong>null</strong><a name="558" href="#558">558</a> && registeredCrawlURIDispositionListeners.size() > 0) {<a name="559" href="#559">559</a> Iterator it = registeredCrawlURIDispositionListeners.iterator();
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -