https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246833#M57 最近作了測試<BR /><BR />B node Add oracle Data file,由速度非常慢<BR /><BR />add 1G datafile 將近30 minutes<BR /><BR /><BR /><BR /><BR /><BR />create second datafile 的時候,做到一半我就用cfsmgr切換到A node,<BR /><BR />結果不到10 seconds 就結束了,所以說應該是整個I/O尚未結速就會切換動作<BR /><BR /> Wed, 19 Apr 2006 04:02:25 GMT watermelonyu 2006-04-19T04:02:25Z https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246831#M55 當我做relocate mount point 從node1 到node2的時候<BR /><BR />example<BR /><BR />cfsmgr -h node1 -r -a SERVER=node2 /u01<BR /><BR />例如我正在複製檔案 and 異動檔案,<BR /><BR />所以i/o都是走node1 才對<BR /><BR />"Because some metadata operations (file size changes) must go to CFS server via function shipping<BR /><BR />"<BR /><BR />這時候用cfsmgr relocate切換的時候會等i/o完畢才切換嗎<BR /><BR /><BR /><BR /><BR /><BR /> Sun, 16 Oct 2005 00:00:03 GMT https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246831#M55 watermelonyu 2005-10-16T00:00:03Z https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246832#M56 你兩個node都有裝collect嗎 ?<BR /><BR />做各測試, 再用collect看就知道囉. <BR /><BR /><BR /><BR />不過實際的情況, 就算你把serv node調開, i/o也不會切的很乾淨. Sun, 16 Oct 2005 00:35:28 GMT https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246832#M56 watermelonyu 2005-10-16T00:35:28Z https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246833#M57 最近作了測試<BR /><BR />B node Add oracle Data file,由速度非常慢<BR /><BR />add 1G datafile 將近30 minutes<BR /><BR /><BR /><BR /><BR /><BR />create second datafile 的時候,做到一半我就用cfsmgr切換到A node,<BR /><BR />結果不到10 seconds 就結束了,所以說應該是整個I/O尚未結速就會切換動作<BR /><BR /> Wed, 19 Apr 2006 04:02:25 GMT https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246833#M57 watermelonyu 2006-04-19T04:02:25Z https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246834#M58 寫錯了,是切到B node Wed, 19 Apr 2006 04:14:30 GMT https://community.hpe.com/t5/trucluster/when-cfsmgr-relocates/m-p/2246834#M58 watermelonyu 2006-04-19T04:14:30Z