On Mon, Jul 22, 2019 at 3:26 PM Ajitha Robert <ajitharober...@gmail.com> wrote: > > Thanks for your reply > > 1) In scenario 1, I didnt attempt to delete the cinder volume. Please find > the cinder volume log. > http://paste.openstack.org/show/754731/
It might be better to ping Cinder folks about that one. It doesn't really make sense to me from a quick glance. > > 2) In scenario 2. I will try with debug. But i m having a test setup with one > OSD in primary and one OSD in secondary. distance between two ceph clusters > is 300 km > > > 3)I have disabled ceph authentication totally for all including rbd-mirror > daemon. Also i have deployed the ceph cluster using ceph-ansible. Will these > both create any issue to the entire setup Not to my knowledge. > 4)The image which was in syncing mode, showed read only status in secondary. Mirrored images are either primary or non-primary. It is the expected (documented) behaviour that non-primary images are read-only. > 5)In a presentation i found as journaling feature is causing poor performance > in IO operations and we can skip the journaling process for mirroring... Is > it possible.. By enabling mirroring to entire cinder pool as pool mode > instead of mirror mode of rbd mirroring.. And we can skip the > replication_enabled is true spec in cinder type.. Journaling is required for RBD mirroring. > > > > On Mon, Jul 22, 2019 at 11:13 PM Jason Dillaman <jdill...@redhat.com> wrote: >> >> On Mon, Jul 22, 2019 at 10:49 AM Ajitha Robert <ajitharober...@gmail.com> >> wrote: >> > >> > No error log in rbd-mirroring except some connection timeout came once, >> > Scenario 1: >> > when I create a bootable volume of 100 GB with a glance image.Image get >> > downloaded and from cinder, volume log throws with "volume is busy >> > deleting volume that has snapshot" . Image was enabled with exclusive >> > lock, journaling, layering, object-map, fast-diff and deep-flatten >> > Cinder volume is in error state but the rbd image is created in primary >> > but not in secondary. >> >> Any chance you know where in Cinder that error is being thrown? A >> quick grep of the code doesn't reveal that error message. If the image >> is being synced to the secondary site when you attempt to delete it, >> it's possible you could hit this issue. Providing debug log messages >> from librbd on the Cinder controller might also be helpful for this. >> >> > Scenario 2: >> > but when i create a 50gb volume with another glance image. Volume get >> > created. and in the backend i could see the rbd images both in primary and >> > secondary >> > >> > From rbd mirror image status i found secondary cluster starts copying , >> > and syncing was struck at around 14 %... It will be in 14 % .. no progress >> > at all. should I set any parameters for this like timeout?? >> > >> > I manually checked rbd --cluster primary object-map check <object-name>.. >> > No results came for the objects and the command was in hanging.. Thats why >> > got worried on the failed to map object key log. I couldnt even rebuild >> > the object map. >> >> It sounds like one or more of your primary OSDs are not reachable from >> the secondary site. If you run w/ "debug rbd-mirror = 20" and "debug >> rbd = 20", you should be able to see the last object it attempted to >> copy. From that, you could use "ceph osd map" to figure out the >> primary OSD for that object. >> >> > the image which was in syncing mode, showed read only status in secondary. >> > >> > >> > >> > On Mon, 22 Jul 2019, 17:36 Jason Dillaman, <jdill...@redhat.com> wrote: >> >> >> >> On Sun, Jul 21, 2019 at 8:25 PM Ajitha Robert <ajitharober...@gmail.com> >> >> wrote: >> >> > >> >> > I have a rbd mirroring setup with primary and secondary clusters as >> >> > peers and I have a pool enabled image mode.., In this i created a rbd >> >> > image , enabled with journaling. >> >> > >> >> > But whenever i enable mirroring on the image, I m getting error in >> >> > osd.log. I couldnt trace it out. please guide me to solve this error. >> >> > >> >> > I think initially it worked fine. but after ceph process restart. these >> >> > error coming >> >> > >> >> > >> >> > Secondary.osd.0.log >> >> > >> >> > 2019-07-22 05:36:17.371771 7ffbaa0e9700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/journal/cls_journal.cc:61: failed to get >> >> > omap key: client_a5c76849-ba16-480a-a96b-ebfdb7f6ac65 >> >> > 2019-07-22 05:36:17.388552 7ffbaa0e9700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/journal/cls_journal.cc:472: active object >> >> > set earlier than minimum: 0 < 1 >> >> > 2019-07-22 05:36:17.413102 7ffbaa0e9700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/journal/cls_journal.cc:61: failed to get >> >> > omap key: order >> >> > 2019-07-22 05:36:23.341490 7ffbab8ec700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/rbd/cls_rbd.cc:4125: error retrieving image >> >> > id for global id '9e36b9f8-238e-4a54-a055-19b19447855e': (2) No such >> >> > file or directory >> >> > >> >> > >> >> > primary-osd.0.log >> >> > >> >> > 2019-07-22 05:16:49.287769 7fae12db1700 0 log_channel(cluster) log >> >> > [DBG] : 1.b deep-scrub ok >> >> > 2019-07-22 05:16:54.078698 7fae125b0700 0 log_channel(cluster) log >> >> > [DBG] : 1.1b scrub starts >> >> > 2019-07-22 05:16:54.293839 7fae125b0700 0 log_channel(cluster) log >> >> > [DBG] : 1.1b scrub ok >> >> > 2019-07-22 05:17:04.055277 7fae12db1700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/journal/cls_journal.cc:472: active object >> >> > set earlier than minimum: 0 < 1 >> >> > >> >> > 2019-07-22 05:33:21.540986 7fae135b2700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/journal/cls_journal.cc:472: active object >> >> > set earlier than minimum: 0 < 1 >> >> > 2019-07-22 05:35:27.447820 7fae12db1700 0 <cls> >> >> > /build/ceph-12.2.12/src/cls/rbd/cls_rbd.cc:4125: error retrieving image >> >> > id for global id '8a61f694-f650-4ba1-b768-c5e7629ad2e0': (2) No such >> >> > file or directory >> >> >> >> Those don't look like errors, but the log level should probably be >> >> reduced for those OSD cls methods. If you look at your rbd-mirror >> >> daemon log, do you see any errors? That would be the important place >> >> to look. >> >> >> >> > >> >> > -- >> >> > Regards, >> >> > Ajitha R >> >> > _______________________________________________ >> >> > ceph-users mailing list >> >> > ceph-users@lists.ceph.com >> >> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >> >> >> >> >> >> >> >> -- >> >> Jason >> >> >> >> -- >> Jason > > > > -- > Regards, > Ajitha R -- Jason _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com