Hi Jonathan: "the old node can be the replacement, as long as you change its IP address"
Do you mean that the operations to replace a bad node is : 1) choose a new machine which has the same configuration, eg. InitialToken, and has a different IP address; 2) start the new machine, which will start boostrapping; 3) After bootstrapping, the new machine will restore the data as before. (All nodes' InitialToken are set manully) I have tried in this way and that looks ok. Is this a good way? :-) Thanks !! ------------------ XL.Pan 2010-01-18 ------------------------------------------------------------- 发件人：Jonathan Ellis 发送日期：2010-01-15 11:12:05 收件人：cassandra-user 抄送： 主题：Re: Re: replace a bad node through bootstrapping On Thu, Jan 14, 2010 at 9:02 PM, Michael Lee <mail.list.steel.men...@gmail.com> wrote: > If a node's data has been damaged, you cannot use new node replace old one > directly, unless 'removetoken' first. > > But, (suppose node A is dead) > 'removetoken' will complement missing replica due A's death first, it will > generate lot data on other nodes, say it's B, C, D > After add new node and copy data from other node through bootstrapping, you > have to 'cleanup' data just > generate from ' removetoken ' on B, C, D > > So, B/C/D will have heavy I/O load (half of them is waste) due to repair A, > in pan's case, it will be 5TB (and will cause days...) > > Pan try to invent a method to repair A directly through streaming, and have > less impact on other nodes. Thanks for clarifying that. I thought we agreed in your last thread about this that bootstrapping a replacement node (the old node can be the replacement, as long as you change its IP address) first, then removing the entry for the dead one, would be a reasonable procedure here. -Jonathan
