Hi Kouya, Thank you for reporting and trying to fix this! Could you please create a tracker ticket [1] to make sure it is not lost? And I think it would be much easier to review your patches (and bring the core developers attention) if you created a pull request [2]. And if you do this please add the PR link to the tracker ticket. If you have problems with any of this just let us know, I can do it for you. [1] http://tracker.ceph.com/projects/rados [2] https://github.com/ceph/ceph On Mon, May 28, 2018 at 06:36:18PM +0900, Kouya Shimura wrote: > Hi, > > I've found that a PG is eternally stuck in 'unfound_recovery' after > some OSDs are marked down. > > For example, the following steps reproduce this. > > 1) Create EC 2+1 pool. Assume a PG has [1,0,2] up/acting set. > 2) Execute "ceph osd out osd.0 osd.2". Now the PG has [1,3,5] up/acting set. > 3) Put some objects to the PG. > 4) Execute "ceph osd in osd.0 osd.2". It starts recovering to [1,0,2]. > 5) Execute "ceph osd down osd.3 osd.5". (These downs are fake. osd.3 > and osd.5 are actually not down) > It leads the PG to transit 'unfound_recovery' and stay on forever. > > Interestingly, this bad situation is resolved by mean of marking down > another OSD. > > 6) Executing "ceph osd down osd.0" (any OSD in acting set is ok) resolves > 'unfound_recovery' and restart recovering. > > > Upon my investigation, if downed OSD is not a member of current up/acting set, > a PG might stay 'ReplicaActive' and discard peering requests from the primary. > Thus the primary OSD can't exit from unfound state. > PGs of downed OSD should transit to 'Reset' state and start peering. > > > I'll post two patches. The first one fixes this issue. > The second one is trivial optimization (optional). > > Thanks, > Kouya > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Mykola Golub -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
