Ceph : Issueshttps://tracker.ceph.com/https://tracker.ceph.com/favicon.ico2010-04-16T11:22:30ZCeph
Redmine CephFS - Bug #46 (Closed): Remove asker from MDiscoverhttps://tracker.ceph.com/issues/462010-04-16T11:22:30ZGreg Farnumgfarnum@redhat.com
<p>It's redundant and confusing with get_orig_source(); and it's unecessary since MDiscover messages aren't forwarded anymore.</p>
<p>When done, switch handle_discover to use a Connection-based send_message</p> CephFS - Feature #45 (Closed): Investigate adding Connection * to mds Sessionhttps://tracker.ceph.com/issues/452010-04-16T11:21:11ZGreg Farnumgfarnum@redhat.com
<p>This would open up several more uses of connection-based send_message.</p> CephFS - Bug #44 (Closed): Make MDS use new connection-based send_messagehttps://tracker.ceph.com/issues/442010-04-15T16:42:42ZGreg Farnumgfarnum@redhat.com
<p>This will be more complicated than the OSD adjustments because there are a lot of send_message_[client|mon|mds|osd] calls.</p> Ceph - Feature #43 (Resolved): Make OSD use new connection-based https://tracker.ceph.com/issues/432010-04-15T16:39:10ZGreg Farnumgfarnum@redhat.com
<p>We have a new send_message(Message *m, Connection *c); let's use it in the OSD!</p> Ceph - Bug #39 (Closed): Re-examine PaxosServiceMessage::get_session() and usehttps://tracker.ceph.com/issues/392010-04-14T16:56:49ZGreg Farnumgfarnum@redhat.com
<p>I got a crash while testing the new libatomic-based atomic_t, but I don't think it could have been caused by the change. Segfault on a null-valued Session.</p> rgw - Bug #31 (Closed): Handle mal-formed requests betterhttps://tracker.ceph.com/issues/312010-04-13T13:21:35ZGreg Farnumgfarnum@redhat.com
<p>RGW can't handle requests that don't include things like the HTTP_HOST header -- it just crashes out.</p>
<p>This was encountered when trying to test Cyberduck with the service; it can't connect. I don't know if Cyberduck is creating malformed requests or if rgw can't accept valid ones, but we probably shouldn't crash out on NPEs either way!</p> CephFS - Feature #13 (Resolved): uclient: Make readdir use the cachehttps://tracker.ceph.com/issues/132010-04-09T13:07:39ZGreg Farnumgfarnum@redhat.com
<p>Right now, readdir doesn't actually read out of the cache but sends a request to the MDS on every activation. This is slow, and should be changed so that it uses the cache appropriately.</p> CephFS - Feature #12 (Resolved): uclient: Make cap handling smarterhttps://tracker.ceph.com/issues/122010-04-09T13:04:21ZGreg Farnumgfarnum@redhat.com
<p>Right now the uclient isn't very intelligent about dropping capabilities. Change the cap drops associated with unlink, rmdir, and rename so they more closely mimic the kclient.</p> CephFS - Cleanup #11 (Resolved): mds: replace ALLOW_MESSAGES_FROM macrohttps://tracker.ceph.com/issues/112010-04-09T12:56:35ZGreg Farnumgfarnum@redhat.com
<p>This is in MDS::_dispatch. This macro is used as a basic permission checker, but any permissions should be dealt with through the caps infrastructure.</p>
<p>This will require actually implementing some MDS caps.</p> Ceph - Feature #10 (Resolved): osd: Replace ALLOW_MESSAGES_FROM macrohttps://tracker.ceph.com/issues/102010-04-09T12:55:05ZGreg Farnumgfarnum@redhat.com
<p>This is in OSD::_dispatch. This macro is used as a basic permission checker, but any permissions should be dealt with through OSDCaps.</p>
<p>If necessary, enhance the system's caps data members and functions so this makes sense and is effective.</p> rgw - Feature #9 (Won't Fix): Access unimported datahttps://tracker.ceph.com/issues/92010-04-09T12:50:41ZGreg Farnumgfarnum@redhat.com
<p>Right now, rgw can only access data which has been added through rgw, even if a user's auid is set in rgw and matches that used on the data.<br />Due to architectural issues with rgw, fixing this may be a bit messy, but if demand comes up rgw should be able to auto-generate default ACLs and allow access to data which a user is known to own (both pools and objects).</p> rgw - Support #8 (Closed): Document differences from S3https://tracker.ceph.com/issues/82010-04-09T12:46:03ZGreg Farnumgfarnum@redhat.com
<p>RGW is a pretty good implementation, but it does have some differences from the S3 api. Document these.</p>
<p>They include: lack of S3's new versioning, no email-based ACLs (though the required framework is largely in-place), all users getting anon_user permissions, and rgw doesn't watch for replayed operations (this is an S3 security measure to prevent attacks on somebody's secret ID).</p> Ceph - Documentation #7 (Resolved): Document Monitor Commandshttps://tracker.ceph.com/issues/72010-04-09T12:38:39ZGreg Farnumgfarnum@redhat.com
<p>There's no centralized area to look for what monitor commands are available, or what permissions they require. (Of course, right now they just all require admin, but that could change.)</p> Ceph - Feature #6 (Rejected): libceph could use a backward-compatible-to functionhttps://tracker.ceph.com/issues/62010-04-09T12:36:17ZGreg Farnumgfarnum@redhat.com
<p>If this library is revised it'd be nice to have a way to check compatibility between versions.</p> Ceph - Bug #5 (Closed): ./rados lspools sometimes hangs after listing all pools?https://tracker.ceph.com/issues/52010-04-09T12:35:23ZGreg Farnumgfarnum@redhat.com
<p>This is rare and intermittent, but happens occasionally.<br />Best guess so far is that this locks the OSDMap but doesn't explicitly wake up before exiting so if an OSDMap update comes in that thread never wakes and the program hangs.</p>