On Thu, Oct 20, 2011 at 7:07 AM, 吕鹏 <[email protected]> wrote: > Hello everyone: > My hbase&hadoop is cdh3u0.
Why not run u1? > To test the cluster, i use 10 client to write > date to hbase concurrently. I implement the client by hbase java api. When > the test run for several hours, the 10 clients are all crashed, and two > kinds of exceptions is catched. How can i solve the problem? (I > set hbase.hregion.max.filesize to 2G, and do not pre-sharding hbase table. > Can this cause the problem?) You should presplit. > Exception in thread "main" > org.apache.hadoop.hbase.client.RegionOfflineException: region offline: > lzo10DB,\x89\xB7\x17\xD2\x08\xA9\x87\xFFg\xE2\x9Fxy8\xD0\x8A\r\xA7\x96\xB9\x9D|D\x8D\xE9\xB1\x8B\x94\xDF5\xA3\xDD\x86O\x7F#f%S\x93\xE0F\x1E|\x07\xF9Q\xF3\xD4,1319046253246.dbc7a6366cdd86019af7d56284dc5d77. The region is offline. Either it took too long for it to be put back on line after a server crash or a split or its not online at all. Grep around in your master log to see if you can figure what happened with this region. > Exception in thread "main" > org.apache.hadoop.hbase.client.RetriesExhaustedWithDetailsException: Failed > 1 action: NotServingRegionException: 1 time, servers with issues: > sds-179-36:60020, Client thinks region is on a server but server says it doesn't have it. Again grep around in master log to try see what happened. Likely, the accounting for this region got messed up. St.Ack
