The source node going down (whether it was really down/unavailable or a FD false positive doesn't matter) will prevent bootstrap from completing. You'll have to restart the new node.
On Wed, Sep 15, 2010 at 5:09 AM, Matthias L. Jugel <l...@thinkberg.com> wrote: > Hi, > > I would like to extend our cassandra system by another node. It contains > about 800GB data right now. I have set up another host and installed > cassandra (both 0.6.3). The new cassandra node starts up and contacts the > first to receive "load info" and then tries to receive the bootstrap token: > > DEBUG 11:35:12,277 attempting to connect to /10.0.1.1 > INFO 11:35:12,975 Node /10.0.1.1 is now part of the cluster > DEBUG 11:35:12,976 Node /10.0.1.1 state normal, token mLsdpDaEx6qzriO9 > DEBUG 11:35:12,977 No bootstrapping or leaving nodes -> empty pending ranges > for TEST > DEBUG 11:35:12,977 No bootstrapping or leaving nodes -> empty pending ranges > for Twimpact > DEBUG 11:35:13,276 Disseminating load info ... > INFO 11:35:13,977 InetAddress /10.0.1.1 is now UP > INFO 11:35:13,978 Started hinted handoff for endPoint /10.0.1.1 > INFO 11:35:13,984 Finished hinted handoff of 0 rows to endpoint /10.0.1.1 > DEBUG 11:36:13,276 Disseminating load info ...DEBUG 11:36:41,278 ... got load > info > INFO 11:36:41,278 Joining: getting bootstrap token > DEBUG 11:36:41,281 attempting to connect to /10.0.1.1 > DEBUG 11:37:13,276 Disseminating load info ... > INFO 11:37:55,273 InetAddress /10.0.1.1 is now dead. > DEBUG 11:38:13,276 Disseminating load info ... > INFO 11:38:29,603 InetAddress /10.0.1.1 is now UP > INFO 11:38:29,603 Started hinted handoff for endPoint /10.0.1.1 > INFO 11:38:29,604 Finished hinted handoff of 0 rows to endpoint /10.0.1.1 > > nodetool on this new host gives me: > > ./bin/nodetool -h localhost streams > Mode: Joining: getting bootstrap token > Not sending any streams. > Not receiving any streams. > > I'd like to know how long it can take until the bootstrap token is received. > I guess we may have about 500 million keys on the seed node. > What I was wondering too, is the short host is dead and host is UP notice. I > have no idea where this comes from. The seed node says operation mode is > normal. > > Thanks, > > Leo. > -- > Matthias L. Jugel - Idea Transporter > Fon: +49 170 226 1897, Fax: +49 30 91685761 > http://thinkberg.com | http://twitter.com/thinkberg > > There are two kinds of fool, one says "this is old and therefore good" > and the other says "this is new and therefore better". Philip K. Dick. > > -- Jonathan Ellis Project Chair, Apache Cassandra co-founder of Riptano, the source for professional Cassandra support http://riptano.com
