Can you describe what is happening in more detail? How many ZK servers?
Do they have dedicated log and snapshot spindles? Do you have log files? Why do you think data might be lost? Do you have any evidence of this or are you just worried? Why do you think that the down server can't join the cluster? Is this something you observed? If you have seen a problem, please put your log files somewhere that people can look at them. Here are some links: http://www.gerv.net/hacking/how-to-ask-good-questions/ http://wiki.apache.org/hadoop/ZooKeeper/Troubleshooting 2011/6/9 yaoxinming <[email protected]>: > hi all： > when a zookeeper server crash down sometimes and then up , > at that time, the system is under high write (5k/s),maybe it is hard > to recover the data and some data maybe lost, > I think the down server can't join the cluster. > > > thanks for any advice > > 2011.6.9 > xinming.yao >
