Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
Zookeeper >> mail # user >> Zookeeper ensemble backup questions?


+
jack ma 2013-07-19, 18:00
+
Sergey Maslyakov 2013-07-19, 18:19
+
jack ma 2013-07-19, 18:42
+
Sergey Maslyakov 2013-07-19, 19:15
+
Sergey Maslyakov 2013-07-19, 19:29
+
jack ma 2013-07-19, 22:28
+
Thawan Kooburat 2013-07-19, 21:24
+
jack ma 2013-07-19, 22:32
+
Thawan Kooburat 2013-07-19, 22:49
+
Jordan Zimmerman 2013-07-19, 18:13
+
Sergey Maslyakov 2013-07-19, 18:21
+
Alexander Shraer 2013-07-19, 18:26
+
Sergey Maslyakov 2013-07-19, 18:38
Copy link to this message
-
Re: Zookeeper ensemble backup questions?
thanks, Sergey!

On Fri, Jul 19, 2013 at 11:38 AM, Sergey Maslyakov <[EMAIL PROTECTED]> wrote:
> Alexander,
>
> I cannot speak to that intelligently. I don't know implementation and
> design details well enough. However, a snapshot of this kind, if it is
> produced by ZooKeeperServer.takeSnapshot() method, should contain not just
> the DataTree content but the sessions too. I am not sure if this covers
> ephemeral nodes though but I am pretty sure this is way Zookeeper server
> takes snapshots driven by the transaction count.
>
>
> Regards,
> /Sergey
>
>
> On Fri, Jul 19, 2013 at 1:26 PM, Alexander Shraer <[EMAIL PROTECTED]> wrote:
>
>> Sergey,
>>
>> Can you explain what happens with ephemeral nodes during snapshot and
>> restore ?
>>
>> Thanks,
>> Alex
>>
>> On Fri, Jul 19, 2013 at 11:21 AM, Sergey Maslyakov <[EMAIL PROTECTED]>
>> wrote:
>> > Jordan,
>> >
>> > It depends on the use case. Some use cases deal with 99% or more of
>> static
>> > non-ephemeral nodes. And this is why the Exhibitor approach does not work
>> > for those use cases.
>> >
>> >
>> > Regards,
>> > /Sergey
>> >
>> >
>> > On Fri, Jul 19, 2013 at 1:13 PM, Jordan Zimmerman <
>> > [EMAIL PROTECTED]> wrote:
>> >
>> >> I'm a bit late to this one but I don't understand the need for complete
>> >> backup of ZK data. In my experience, 99% of ZNodes are ephemeral.
>> >> Therefore, it would be wrong to restore those nodes. In a disaster, the
>> >> connection sessions would expire and you would not want ephemeral nodes
>> >> restored. This is why in Exhibitor I took the approach of selective
>> restore.
>> >>
>> >> -Jordan
>> >>
>> >> On Jul 19, 2013, at 11:00 AM, jack ma <[EMAIL PROTECTED]> wrote:
>> >>
>> >> > I asked those question in the thread
>> >> >
>> >>
>> http://mail-archives.apache.org/mod_mbox/zookeeper-user/201307.mbox/%3cCAB+cfdwhOV0JfB04=MpO_+i-4ou=VbL=EG2XS557+j+[EMAIL PROTECTED]%3e
>> >> ,
>> >> > but there is no response for that.
>> >> >
>> >> > So I posted those questions again here, hopefully I could get helps
>> >> > from the community.
>> >> >
>> >> > I want to make sure I am fully understanding the procedures of
>> zookeeper
>> >> > backup and disaster recovery:
>> >> >
>> >> > For the backup procedures at zookeeper assemble:
>> >> > (1) Login to any host which state is "Serving"
>> >> >           Question:
>> >> >                  Do I have to login to leader node, or any node is ok?
>> >> > (2) Copy latest snapshot file and transaction log from version-2
>> >> directory.
>> >> >           Question:
>> >> >                  How to make sure we do not copy corrupt files if the
>> >> > snapshot/transaction log is in the middle of update? Do we have to
>> >> shutdown
>> >> > the node to make the copy?
>> >> >                  besides the transaction log and snapshot, do we have
>> to
>> >> > copy other files such as the ecoch files
>> >> >
>> >> > For the disaster recovery procedures at zookeeper assemble:
>> >> > (1) recreate the machines for the zookeeper ensemble
>> >> > (2) copy snapshot/transaction log we backed up into the zookeeper
>> >> > dataDir\version-2 and logDir\version2.
>> >> >           Question:
>> >> >                 Do we have to copy the epoch files?
>> >> >                 Do we have to copy snapshot/transaction log backed up
>> to
>> >> > all the zookeeper node, or just the first node we starts?
>> >> >
>> >> > Appreciate your time and help.
>> >> > Jack
>> >>
>> >>
>>