[ceph-users] rbd-target-api service fails to start with address family not supported

Jason Dillaman jdillama at redhat.com
Mon Mar 18 08:19:09 PDT 2019


Looks like you have the IPv6 stack disabled. You will need to override
the bind address from "[::]" to "0.0.0.0" via the "api_host" setting
[1] in "/etc/ceph/iscsi-gateway.cfg"

[1] https://github.com/ceph/ceph-iscsi/blob/master/ceph_iscsi_config/settings.py#L100

On Mon, Mar 18, 2019 at 11:09 AM Wesley Dillingham
<wdillingham at godaddy.com> wrote:
>
> I am having some difficulties getting the iscsi gateway and api setup. Working with a 12.2.8 cluster. And the gateways are Centos 7.6.1810 kernel 3.10.0-957.5.1.el7.x86_64
>
> Using a previous version of ceph iscsi packages:
> ceph-iscsi-config-2.6-2.6.el7.noarch
> ceph-iscsi-tools-2.1-2.1.el7.noarch
> ceph-iscsi-cli-2.7-2.7.el7.noarch
>
> I encountered the error when attemping to create my rbd device:
>
> /disks> create rbd image=iscsi_poc_1 size=500g max_data_area_mb=8
> Failed : 500 INTERNAL SERVER ERROR
>
>
> This created the RBD but when attempting to list the iscsi disks in gwcli nothing is displayed/registered. I consulted the mailing list etc and the first suggestion was to ensure tcmu runner is running and it is. Also I am running the latest tcmu-runner and libtcmu as acquired from shaman.
>
> Some further mailing reading suggested I run a more recent version of the ceph iscsi packages and so I upgraded to the latest bulds from shaman:
> ceph-iscsi-config-2.6-80.g24deeb2.el7.noarch
> ceph-iscsi-cli-2.7-105.g4802654.el7.noarch
>
> This gives me the following errors from journalctl:
>
> ]:  * Running on http://[::]:5000/
> ]: Traceback (most recent call last):
> ]: File "/usr/bin/rbd-target-api", line 2053, in <module>
> ]: main()
> ]: File "/usr/bin/rbd-target-api", line 2002, in main
> ]: ssl_context=context)
> ]: File "/usr/lib/python2.
> ]: run_simple(host, port, self, **options)
> ]: File "/usr/lib/python2.
> ]: inner()
> ]: File "/usr/lib/python2.
> ]: passthrough_errors, ssl_context).serve_forever()
> ]: File "/usr/lib/python2.
> ]: passthrough_errors, ssl_context)
> ]: File "/usr/lib/python2.
> ]: HTTPServer.__init__(self, (host, int(port)), handler)
> ]: File "/usr/lib64/python2.
> ]: self.socket_type)
> ]: File "/usr/lib64/python2.
> ]: _sock = _realsocket(family, type, proto)
> socket.error: [Errno 97] Address family not supported by protocol
>
> my iscsi-gateway.conf is as follows:
>
> [config]
> cluster_name = ceph
> gateway_keyring = ceph.client.admin.keyring
> api_secure = false
> trusted_ip_list = <list of both gateway ipv4 in a comma separated list>
> prometheus_exporter = false
> debug = true
>
> Firewalls are disabled and both nodes can speak to eachother on 5000
> the gw service is able to be started independently however the api service is busted.
>
> I am hoping someone can advise on how to get past my most recent: "socket.error: [Errno 97] Address family not supported by protocol" and generally recommend the most stable current iteration of rbd iscsi to be used with a luminous or later cluster.
>
> Thank you.
>
>
> Respectfully,
>
> Wes Dillingham
> wdillingham at godaddy.com
> Site Reliability Engineer IV - Platform Storage / Ceph
>
> _______________________________________________
> ceph-users mailing list
> ceph-users at lists.ceph.com
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com



-- 
Jason


More information about the ceph-users mailing list