Hi,
I wanted to understand the nature of the RGW Threads Being Blocked on Requests for a PG which is currently in INACTIVE State
1.As long as the PG is inactive the requests stay blocked
2.Could the RGW Threads Use Event Based Model, if a PG is inactive, put the Current Request into a Block Queue, A event based Model, which is something similar to nginx
3.Could The RGW threads Timeout, if the Request stay blocked then a certain threshold?
4.Was the Design of Blocking the RGW threads on a Inactive PG by choice, or it was they way this model was implemented?
5.Are there any serialisation issues that could arise if a async model is used?
The above questions are based upon the observation seen on Hammer, and the reason for the above is to Increase the Availability of a WebStack service, is slight percentages of PG are down for a increased amount of time.
Thanks
Romit
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com