Re: [ceph-users] "STATE_CONNECTING_WAIT_BANNER_AND_IDENTIFY" showing in ceph -s

[nokia ceph]

Thanks Greg for the info.

As per our testing, we fix this warning problem by disabling ceph-mgr
service on all the ceph nodes .if the warning still persist, we go  on the
last ceph node of the cluster and  tried starting and stoping ceph-mgr
service as this operation solved the issue.

Do you have any other suggestion to how to skip this warning?

Thanks


On Mon, Feb 27, 2017 at 8:47 PM, Gregory Farnum  wrote:

> On Sun, Feb 26, 2017 at 10:41 PM, nokia ceph 
> wrote:
> > Hello,
> >
> > On a fresh installation ceph kraken 11.2.0 , we are facing below error in
> > the "ceph -s" output.
> >
> > 
> > 0 -- 10.50.62.152:0/675868622 >> 10.50.62.152:6866/13884
> conn(0x7f576c002750
> > :-1 s=STATE_CONNECTING_WAIT_BANNER_AND_IDENTIFY pgs=0 cs=0
> > l=1)._process_connection connect claims to be 10.50.62.152:6866/1244305
> not
> > 10.50.62.152:6866/13884 - wrong node!
> > 
>
> As you see when comparing addresses, they differ only at the end, in
> what we call the nonce. This most commonly just means that one end or
> the other has a newer osd map epoch indicating the OSD went down and
> it restarted itself. If it persists once they've all finished their
> startup work, you may have an issue with your network config or
> something.
> -Greg
>
> >
> > May I know under what scenerio the above message will prompt in the
> screen.
> > Also let me know what is the impact of this message.
> >
> > I suspect this message raised because of something wrong with the OSD
> > creation.
> >
> > Env:-
> > Kraken - 11.2.0 , 4 node , 3 mon
> > RHEL 7.2
> > EC 3+1 , 68 disks , bluestore
> >
> > Please suggest how to remove or skip these errors.
> > FYI -
> > https://github.com/ceph/ceph/blob/master/src/msg/async/
> AsyncConnection.h#L237
> >
> > Thanks
> >
> >
> > ___
> > ceph-users mailing list
> > ceph-users@lists.ceph.com
> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
> >
>
___
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Re: [ceph-users] "STATE_CONNECTING_WAIT_BANNER_AND_IDENTIFY" showing in ceph -s

[Gregory Farnum]

On Sun, Feb 26, 2017 at 10:41 PM, nokia ceph  wrote:
> Hello,
>
> On a fresh installation ceph kraken 11.2.0 , we are facing below error in
> the "ceph -s" output.
>
> 
> 0 -- 10.50.62.152:0/675868622 >> 10.50.62.152:6866/13884 conn(0x7f576c002750
> :-1 s=STATE_CONNECTING_WAIT_BANNER_AND_IDENTIFY pgs=0 cs=0
> l=1)._process_connection connect claims to be 10.50.62.152:6866/1244305 not
> 10.50.62.152:6866/13884 - wrong node!
> 

As you see when comparing addresses, they differ only at the end, in
what we call the nonce. This most commonly just means that one end or
the other has a newer osd map epoch indicating the OSD went down and
it restarted itself. If it persists once they've all finished their
startup work, you may have an issue with your network config or
something.
-Greg

>
> May I know under what scenerio the above message will prompt in the screen.
> Also let me know what is the impact of this message.
>
> I suspect this message raised because of something wrong with the OSD
> creation.
>
> Env:-
> Kraken - 11.2.0 , 4 node , 3 mon
> RHEL 7.2
> EC 3+1 , 68 disks , bluestore
>
> Please suggest how to remove or skip these errors.
> FYI -
> https://github.com/ceph/ceph/blob/master/src/msg/async/AsyncConnection.h#L237
>
> Thanks
>
>
> ___
> ceph-users mailing list
> ceph-users@lists.ceph.com
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
___
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com