hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Yu <yuzhih...@gmail.com>
Subject Re: if/when zookeeper goes down, how to recover?
Date Sat, 14 Jun 2014 21:56:38 GMT
Cycling old bits:

http://search-hadoop.com/m/WsPol11E4ss1/Zookeeper+ec2&subj=Best+practices+for+using+zookeeper+in+an+Amazon+EC2+environment

On Jun 14, 2014, at 2:10 PM, S Ahmed <sahmed1020@gmail.com> wrote:

> Are there any 'best practices' when running zookeeper on ec2?  i.e. the
> latency could be poor and hostname/ip addresses can also change when bring
> a node up/down.
> 
> 
> On Sat, Jun 14, 2014 at 10:13 AM, Ted Yu <yuzhihong@gmail.com> wrote:
> 
>> bq. what is the process do recover?
>> 
>> You need to bring zookeeper nodes back as soon as possible.
>> 
>> bq. will all clients reconnect or do you have to restart all servers?
>> 
>> This depends on the release of hbase you're using and the duration which
>> zookeeper ensemble is down.
>> If the duration is longer than the value for zookeeper.session.timeout,
>> region servers would go down.
>> 
>> See also http://hbase.apache.org/book.html#zookeeper
>> 
>> 
>> On Sat, Jun 14, 2014 at 6:31 AM, S Ahmed <sahmed1020@gmail.com> wrote:
>> 
>>> Hello,
>>> 
>>> If zookeeper goes down (all of them if you have 3/5 nodes), what is the
>>> process do recover?
>>> 
>>> If you simply bring back the zookeeper nodes, will all clients reconnect
>> or
>>> do you have to restart all servers?
>>> 
>>> Thanks.
>> 

Mime
  • Unnamed multipart/alternative (inline, 7-Bit, 0 bytes)
View raw message