Hello, if you want to use multiple IP networks, just use IP routing. Today's switches support such a feature in line rate with great performance. Please keep your firewall/security layer in mind when you use routed networks. I don't believe it would be possible to use any other proxy / modification / ... technology to connect as a Ceph client to a Ceph storage within a different IP network. -- Martin Verges Managing director Mobile: +49 174 9335695 E-Mail: martin.verges@xxxxxxxx Chat: https://t.me/MartinVerges croit GmbH, Freseniusstr. 31h, 81247 Munich CEO: Martin Verges - VAT-ID: DE310638492 Com. register: Amtsgericht Munich HRB 231263 Web: https://croit.io YouTube: https://goo.gl/PGE1Bx Am Fr., 22. März 2019 um 11:32 Uhr schrieb Andres Rojas Guerrero <a.rojas@xxxxxxx>: > > Hi, thank's for the answer, we have seen that the client only have OSD > map for first public network ... > > # cat > /sys/kernel/debug/ceph/88f62260-b8de-499f-b6fe-5eb66a967083.client360894/osdmap > > > you say that the cepfs clients have the same view of the cluster that > have de MON's, that's mean that MON's doesn't have the right vision of > the OSD from the second public network? > > > > > On 22/3/19 9:54, Burkhard Linke wrote: > > Hi, > > > > > > just my 2 ct: > > > > > > Clients do not access the MDS directly, e.g. you do not use the IP > > address of a MDS in the mount command. > > > > > > Clients contact the MONs and retrieve the MDS map, which contains all > > filesystems and their corresponding MDS server(s). The same is also true > > for the OSDs. The clients thus use the MONs' view of the cluster. I'm > > not familar with multiple IP address setups, but I would bet that a > > client is not able to select the correct public network. > > > > > > You can check your setup on the OSD/MDS hosts (are the OSDs/MDSs bound > > to the second network at all?), and in /sys/kernel/debug/ceph/<ceph > > client id>. The later contains the current versions of the MON map, MDS > > map and OSD map. These are the settings used by the client to contact > > the corresponding daemon (assuming kernel cephfs client, ceph-fuse is > > different). > > > > > > Regards, > > > > Burkhard > > > > -- > ******************************************************* > Andrés Rojas Guerrero > Unidad Sistemas Linux > Area Arquitectura Tecnológica > Secretaría General Adjunta de Informática > Consejo Superior de Investigaciones Científicas (CSIC) > Pinar 19 > 28006 - Madrid > Tel: +34 915680059 -- Ext. 990059 > email: a.rojas@xxxxxxx > ID comunicate.csic.es: @50852720l:matrix.csic.es > ******************************************************* > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |