Yes I did. Starting from the scracth(wiped my whole partition off) and playing around with the cache-size I now see good improvement.
With, more search with more threads, Still the search takes 3 - 10 sec. average. Which is far better than what it was before. Still trying with other possibilities. Siva. -----Original Message----- From: Thomas Muller [mailto:[EMAIL PROTECTED] On Behalf Of Thomas Muller Sent: 25 January 2008 14:19 To: 'Apache Directory Developers List' Subject: RE: Searching takes unrealistic time. Did you try what this guy said? Increasing cache-size? | -----Original Message----- | From: Siva Kumar [mailto:[EMAIL PROTECTED] | Sent: 25 January 2008 14:14 | To: Apache Directory Developers List | Subject: RE: Searching takes unrealistic time. | | Thank you for the tip Emmanuel. | | But its not seem to help me. | | Actually, | | I have a custom schema and partition created in ADS and the 40K | entries I mentioned follows the custom objectClass | definition.(below) | | objectClass ( 1.3.6.1.4.1.17370.2.2.1 | NAME 'mbxperson' | SUP top STRUCTURAL | MUST ( uid $ cn $ gn $ sn $ telephoneNumber ) | MAY ( mobileNumber $ ou $ primaryExtention $ title ) | X-ORIGIN 'user defined' ) | | I have attached the server.xml also for your reference. | | My searches are based on names,primaryExtention and telephoneNumber. | The performance looks the same as when no indexes specified. | | I have even wiped the whole data off and recreated the entries again | to make sure the indexes applied correctly. | | Am I missing anything in the loop in this process? Is there any step I | missed out completely?? | | It would be great if you could help? | | Thanks in advance. | | Regards, | Siva. | | | -----Original Message----- | From: Emmanuel Lecharny [mailto:[EMAIL PROTECTED] | Sent: 24 January 2008 16:06 | To: Apache Directory Developers List | Subject: Re: Searching takes unrealistic time. | | Siva Kumar wrote: | > <bean | > | class="org.apache.directory.server.core.partition.impl.btree.MutableIn | > de | > xConfiguration"> | > <property name="attributeId" value="uid" /> | > <property name="cacheSize" value="100" /> | > | Considering you are using the uid attribute, with 40 000 entries in | your base, the current cacheSize will hold 0,25% of the existing uid. | That means you are roughly doing a disk access each time you search | for an entry. | | Increase the cacheSize for each attribute your filters are using. | Here, 40 000 will be ok. | | Be aware that you may have to increase the JVM memory size to avoid | OOM exception too... | | Hope it helps | | -- | -- | cordialement, regards, | Emmanuel Lécharny | www.iktek.com | directory.apache.org | | |
