Re: geo-rep regression because of node-uuid change

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On Tue, Jun 20, 2017 at 4:12 PM, Aravinda <avishwan@xxxxxxxxxx> wrote:
I think following format can be easily adopted by all components

UUIDs of a subvolume are seperated by space and subvolumes are separated by comma

For example, node1 and node2 are replica with U1 and U2 UUIDs respectively and
node3 and node4 are replica with U3 and U4 UUIDs respectively

node-uuid can return "U1 U2,U3 U4"

Geo-rep can split by "," and then split by space and take first UUID
DHT can split the value by space or comma and get unique UUIDs list

Another question is about the behavior when a node is down, existing node-uuid xattr will not return that UUID if a node is down.
After the change [1], if a node is down we send all zeros as the uuid for that node, in the list of node uuids.

[1] https://review.gluster.org/#/c/17084/

Regards,
Karthik
What is the behavior with the proposed xattr?

Let me know your thoughts.

regards
Aravinda VK


On 06/20/2017 03:06 PM, Aravinda wrote:
Hi Xavi,

On 06/20/2017 02:51 PM, Xavier Hernandez wrote:
Hi Aravinda,

On 20/06/17 11:05, Pranith Kumar Karampuri wrote:
Adding more people to get a consensus about this.

On Tue, Jun 20, 2017 at 1:49 PM, Aravinda <avishwan@xxxxxxxxxx
<mailto:avishwan@xxxxxxxxxx>> wrote:


    regards
    Aravinda VK


    On 06/20/2017 01:26 PM, Xavier Hernandez wrote:

        Hi Pranith,

        adding gluster-devel, Kotresh and Aravinda,

        On 20/06/17 09:45, Pranith Kumar Karampuri wrote:



            On Tue, Jun 20, 2017 at 1:12 PM, Xavier Hernandez
            <xhernandez@xxxxxxxxxx <mailto:xhernandez@xxxxxxxxxx>
            <mailto:xhernandez@xxxxxxxxxx
            <mailto:xhernandez@xxxxxxxxxx>>> wrote:

                On 20/06/17 09:31, Pranith Kumar Karampuri wrote:

                    The way geo-replication works is:
                    On each machine, it does getxattr of node-uuid and
            check if its
                    own uuid
                    is present in the list. If it is present then it
            will consider
                    it active
                    otherwise it will be considered passive. With this
            change we are
                    giving
                    all uuids instead of first-up subvolume. So all
            machines think
                    they are
                    ACTIVE which is bad apparently. So that is the
            reason. Even I
                    felt bad
                    that we are doing this change.


                And what about changing the content of node-uuid to
            include some
                sort of hierarchy ?

                for example:

                a single brick:

                NODE(<guid>)

                AFR/EC:

                AFR[2](NODE(<guid>), NODE(<guid>))
                EC[3,1](NODE(<guid>), NODE(<guid>), NODE(<guid>))

                DHT:

                DHT[2](AFR[2](NODE(<guid>), NODE(<guid>)),
            AFR[2](NODE(<guid>),
                NODE(<guid>)))

                This gives a lot of information that can be used to take the
                appropriate decisions.


            I guess that is not backward compatible. Shall I CC
            gluster-devel and
            Kotresh/Aravinda?


        Is the change we did backward compatible ? if we only require
        the first field to be a GUID to support backward compatibility,
        we can use something like this:

    No. But the necessary change can be made to Geo-rep code as well if
    format is changed, Since all these are built/shipped together.

    Geo-rep uses node-id as follows,

    list = listxattr(node-uuid)
    active_node_uuids = list.split(SPACE)
    active_node_flag = True if self.node_id exists in active_node_uuids
    else False

How was this case solved ?

suppose we have three servers and 2 bricks in each server. A replicated volume is created using the following command:

gluster volume create test replica 2 server1:/brick1 server2:/brick1 server2:/brick2 server3:/brick1 server3:/brick1 server1:/brick2

In this case we have three replica-sets:

* server1:/brick1 server2:/brick1
* server2:/brick2 server3:/brick1
* server3:/brick2 server2:/brick2

Old AFR implementation for node-uuid always returned the uuid of the node of the first brick, so in this case we will get the uuid of the three nodes because all of them are the first brick of a replica-set.

Does this mean that with this configuration all nodes are active ? Is this a problem ? Is there any other check to avoid this situation if it's not good ?
Yes all Geo-rep workers will become Active and participate in syncing. Since changelogs will have the same information in replica bricks this will lead to duplicate syncing and consuming network bandwidth.

Node-uuid based Active worker is the default configuration in Geo-rep till now, Geo-rep also has Meta Volume based syncronization for Active worker using lock files.(Can be opted using Geo-rep configuration, with this config node-uuid will not be used)

Kotresh proposed a solution to configure which worker to become Active. This will give more control to Admin to choose Active workers, This will become default configuration from 3.12
https://github.com/gluster/glusterfs/issues/244

--
Aravinda


Xavi




        Bricks:

        <guid>

        AFR/EC:
        <guid>(<guid>, <guid>)

        DHT:
        <guid>(<guid>(<guid>, ...), <guid>(<guid>, ...))

        In this case, AFR and EC would return the same <guid> they
        returned before the patch, but between '(' and ')' they put the
        full list of guid's of all nodes. The first <guid> can be used
        by geo-replication. The list after the first <guid> can be used
        for rebalance.

        Not sure if there's any user of node-uuid above DHT.

        Xavi




                Xavi


                    On Tue, Jun 20, 2017 at 12:46 PM, Xavier Hernandez
                    <xhernandez@xxxxxxxxxx
            <mailto:xhernandez@xxxxxxxxxx> <mailto:xhernandez@xxxxxxxxxx
            <mailto:xhernandez@xxxxxxxxxx>>
                    <mailto:xhernandez@xxxxxxxxxx
            <mailto:xhernandez@xxxxxxxxxx> <mailto:xhernandez@xxxxxxxxxx
            <mailto:xhernandez@xxxxxxxxxx>>>>
                    wrote:

                        Hi Pranith,

                        On 20/06/17 07:53, Pranith Kumar Karampuri wrote:

                            hi Xavi,
                                   We all made the mistake of not
            sending about changing
                            behavior of
                            node-uuid xattr so that rebalance can use
            multiple nodes
                    for doing
                            rebalance. Because of this on geo-rep all
            the workers
                    are becoming
                            active instead of one per EC/AFR subvolume.
            So we are
                            frantically trying
                            to restore the functionality of node-uuid
            and introduce
                    a new
                            xattr for
                            the new behavior. Sunil will be sending out
            a patch for
                    this.


                        Wouldn't it be better to change geo-rep behavior
            to use the
                    new data
                        ? I think it's better as it's now, since it
            gives more
                    information
                        to upper layers so that they can take more
            accurate decisions.

                        Xavi


                            --
                            Pranith





                    --
                    Pranith





            --
            Pranith






--
Pranith




_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-devel

[Index of Archives]     [Gluster Users]     [Ceph Users]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux