fyi fix releases only include fixes. We typically put the new features into minor/major updates. Helps to maintain Operation's sanity. ;-)
Also keep in mind that rolling restart is dead simple and works. 107 will be a nice option to have but is overkill in most scenarios. Patrick On Tue, Nov 29, 2011 at 7:48 PM, Joe Stein <[email protected]> wrote: > 3.4.1 <= pushing the curve =8^) > > /* > Joe Stein > http://www.medialets.com > Twitter: @allthingshadoop > */ > > On Nov 29, 2011, at 10:44 PM, Camille Fournier <[email protected]> wrote: > >> We've been looking at that, see: >> https://issues.apache.org/jira/browse/ZOOKEEPER-107 >> >> It'll take some evaluation though. Hopefully 3.5? >> >> C >> >> On Tue, Nov 29, 2011 at 10:21 PM, nileader <[email protected]> wrote: >>> I personal think zookeeper team need to add the feature of changing >>> zookeeper cluster server list in runtime---without restart server. >>> >>> 2011/11/30 nileader <[email protected]> >>> >>>> Hi，ted，you mean restart old server before new servers? >>>> Let's see the case: >>>> >>>> In my cluster, there are three zk server, A,B and C each config three >>>> server address. >>>> Once, i want to upgrade my cluster to five server. >>>> You say: >>>> >>>>> configure the new servers >>>>> >>>>> - update the configuration on the old servers >>>>> >>>>> - restart each old server one at a time >>>>> >>>>> - start the new servers >>>>> >>>>> >>>> >>>> So, when i config Server C's serverlist, it will be five server in it's >>>> zoo.cfg, and restart it. >>>> Then make change on server b. Once i restart server b, there will be only >>>> two server work in cluster, >>>> and 2< 5/2+1. >>>> >>>> So, how the leader will be elect and how the cluster will work ok? >>>> >>>> >>>> >>>> >>>> >>>> *nileader* ni掌櫃的個人郵箱 >>>> *MSN*： [email protected] >>>> *Weibo*：http://weibo.com/nileader >>>> ------------------------------------------------------------------------------------------------------------------------------------------------ >>>> This email (including any attachments) is confidential and may be legally >>>> privileged, private information of correct recipient and nileader. If you >>>> received this email in error, please delete it immediately and do not copy >>>> it or use it for any purpose or disclose its contents to any other person. >>>> Thank you. >>>> * >>>> 本電郵(包括任何附件)可能含有機密資料並受法律保護，屬於ni掌櫃和正確收件人之間的私有信息。如您不是正確的收件人，請您立即刪除本郵件。請不要將本電郵進行複製並用作任何其它用途，或透露本郵件之內容。謝謝。 >>>> * >>>> >>>> >>>> >>>> 2011/11/29 Ted Dunning <[email protected]> >>>> >>>>> To expand on that FAQ, the detailed process should be to >>>>> >>>>> - configure the new servers >>>>> >>>>> - update the configuration on the old servers >>>>> >>>>> - restart each old server one at a time >>>>> >>>>> - start the new servers >>>>> >>>>> - have your applications reconnect with the full set of servers in the >>>>> connection string. This can be done whenever you like. >>>>> >>>>> Your clients will receive a disconnect message when the server that they >>>>> are talking to is restarted, but they should reconnect immediately and >>>>> transparently. You may have a problem if your applications don't handle >>>>> transient disconnections properly. >>>>> >>>>> >>>>> >>>>> On Tue, Nov 29, 2011 at 3:10 AM, pubudu gunawardena <[email protected] >>>>>> wrote: >>>>> >>>>>> On Tue, Nov 29, 2011 at 1:08 PM, nileader <[email protected]> wrote: >>>>>>> >>>>>>> hi, all. >>>>>>> >>>>>>> If i update my conf/zoo.cfg config, must i restart my zk server? >>>>>>> >>>>>>> In my case, we have 3 servers in production environment, now, we want >>>>> to >>>>>>> upgrade to 5 server. >>>>>> >>>>>> Take a look at http://wiki.apache.org/hadoop/ZooKeeper/FAQ#A6 where it >>>>>> says >>>>>> how to upgrade ZooKeeper. >>>>>>> >>>>>>> >>>>>>> Any problem will happen if a kill one of three servers. >>>>>>> >>>>>>> >>>>>>> *[email protected]* >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> Thanks, >>>>>> Pubudu >>>>>> >>>>> >>>> >>>>
