[ https://issues.apache.org/jira/browse/HBASE-12393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14191953#comment-14191953 ]
stack commented on HBASE-12393: ------------------------------- Why set it to zero? When would it make sense? i.e. loading index blocks every time rather than keeping them around in mem. But yeah, shouldn't NPE. If you have a patch that fixes the NPE and it works for you, attach and we'll commit (but will add WARNINGs that 0 block cache is bad idea as per [~jmspaggi]) > The regionserver web will throw exception if we disable block cache > ------------------------------------------------------------------- > > Key: HBASE-12393 > URL: https://issues.apache.org/jira/browse/HBASE-12393 > Project: HBase > Issue Type: Bug > Components: regionserver > Affects Versions: 0.98.7 > Environment: ubuntu 12.04 64bits, hadoop-2.2.0, hbase-0.98.7-hadoop2 > Reporter: ChiaPing Tsai > Priority: Minor > > The CacheConfig.getBlockCache() will return the null point when we set > hfile.block.cache.size to zero. > It caused the BlockCacheTmplImpl.java:123 to throw null exception. > {code} > org.jamon.escaping.Escaping.HTML.write(org.jamon.emit.StandardEmitter.valueOf(StringUtils.humanReadableInt(cacheConfig.getBlockCache().size())), > jamonWriter); > {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)