On 07/06/2013 04:51 AM, Xue, Chendi wrote:
Hi, all

I wanna fetch debug librbd and debug rbd logs when I am using vm to read / 
write.

Details:
        I created a volume from ceph and attached it to a vm.
        So I suppose when I do read/write in the VM, I can get some rbd debug 
logs in the host where the vm lies in.
        But in fact, after everything I can try, I still cannot get those logs.

Below is the configuration on qemu side
;global
[global]
         ; allow ourselves to open a lot of files
         max open files = 131072
         auth cluster required = none
         auth service required = none
         auth client required = none
         ; set log file
         log file = /var/log/ceph/$name.log
         ; set up pid files
         pid file = /var/run/ceph/$name.pid

;client
[client]
         rbd cache = true
         rbd cache size = 21474836480
         rbd cache max dirty = 0
         ; set qemu log file
         log file = /var/log/ceph/ceph.client.log
#       debug ms = 1
#       debug client = 20
         debug rbd = 20
         debug librbd = 20
         debug objectcacher = 20

; monitors
;  You need at least one.  You need at least three if you want to
;  tolerate any node failures.  Always create an odd number.
[mon]
         mon data = /data/$name
[mon.Ceph-N3]
         host = node3
         mon addr = 192.168.1.13:6789

This configuration is fine.

Something I tried:
        I do can fetch librbd logs when using rbd command like "rbd list pool_name" and 
" rbd -p pool_name bench-write image_name --io-size 4096 --io-threads 1 --io-total 4096 
--io-pattern rand"

But if I using dd in vm to the virtual volume created by ceph, there is still 
no rbd debug logs.

Can the unix user running qemu write to /var/log/ceph to create /var/log/ceph/ceph.client.log?

SELinux or apparmor may be preventing it, if not usual unix permissions.

Josh
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to