On Thu, Aug 22, 2013 at 8:56 AM, christophe courtaut <christophe.courtaut@xxxxxxxxx> wrote: >> On Thu, Aug 22, 2013 at 4:38 AM, christophe courtaut >> <christophe.courtaut@xxxxxxxxx> wrote: >>> Hi Yehuda, >>> >>> I'm currently trying to test the radosgw-agent which lives in >>> https://github.com/ceph/radosgw-agent >>> >>> I modified the vstart.sh script (you can see it here >>> https://github.com/ceph/ceph/pull/522) to be able to launch multiple >>> cluster with vstart.sh in different directories. >>> >>> I created a little script to automate the launch of two clusters and >>> the setup of region and zone, so that i have a master and a slave >>> cluster. >>> >>> With that setup, i encounter two problems: >>> >>> First when i try to create a bucket after setup is done using s3cmd, >>> got this error >>> ERROR: Access to bucket 'TEST' was denied. >>> in the function open_bucket_pool_ctx, it tries to open the pool >>> .rgw.buckets.index, but this pool doesn't seems to exist. >>> Is this behaviour correct? Is it due to my setup? or should this pool >>> be created? >> >> This pool should have been automatically created (if didn't exist >> before). Not sure why it's not happening. > > Ok. I don't know where the automatic creation of this pool should be > handled though. > >>> >>> Second, is that i get a 403 while using the radosgw-agent on the >>> previous setup, with the test.sh script. >>> Am i doing something wrong here? >> >> Did you set the 'system' flag on the appropriate users? > > No i did miss that point. btw the option was not mentionned in the > radosgw-admin help. > here comes a pull request for that : https://github.com/ceph/ceph/pull/528 > >>> To reproduce the first bug: >>> >>> - Get the vstart.sh script from https://github.com/ceph/ceph/pull/522 >>> - Get the script to launch cluster and the *.master *.zone files >>> - Launch the script >>> - Try to create a bucket with s3cmd >>> >>> To reproduce the second one : >>> >>> - Do the same setup as for the first one >>> - Get the radosgw-agent repo >>> - Get the test.sh script in this directory >>> - Launch the test.sh script >>> >>> You can find below the various files i mentionned earlier >>> >>> the script for launching clusters >>> http://pastebin.com/VSJhwRSs >>> >>> master.region file >>> http://pastebin.com/pCrRpM3w >>> >>> master.zone file >>> http://pastebin.com/gi1DDNvr >>> >>> slave.region file >>> http://pastebin.com/idgMDYjj >>> >>> slave.zone file >>> http://pastebin.com/m2KRnqdJ >> >> On both slave and master zone you didn't set the placement pools, so >> it resorts to the default. What does 'radosgw-admin zone get' show on >> both? > > radosgw-admin -c cluster-master/ceph.conf zone get > > { "domain_root": ".rgw", > "control_pool": ".rgw.control", > "gc_pool": ".rgw.gc", > "log_pool": ".log", > "intent_log_pool": ".intent-log", > "usage_log_pool": ".usage", > "user_keys_pool": ".users", > "user_email_pool": ".users.email", > "user_swift_pool": ".users.swift", > "user_uid_pool": ".", > "system_key": { "access_key": "0555b35654ad1656d804", > "secret_key": > "h7GhxuBLTrlhVUyxSPUKUV8r\/2EI4ngqJxD7iBdBYLhwluN30JaT3Q=="}, > "placement_pools": []} > > radosgw-admin -c cluster-slave/ceph.conf zone get > > { "domain_root": ".rgw", > "control_pool": ".rgw.control", > "gc_pool": ".rgw.gc", > "log_pool": ".log", > "intent_log_pool": ".intent-log", > "usage_log_pool": ".usage", > "user_keys_pool": ".users", > "user_email_pool": ".users.email", > "user_swift_pool": ".users.swift", > "user_uid_pool": ".", > "system_key": { "access_key": "0555b35654ad1656d804", > "secret_key": > "h7GhxuBLTrlhVUyxSPUKUV8r\/2EI4ngqJxD7iBdBYLhwluN30JaT3Q=="}, > "placement_pools": []} > > Indeed placement_pools seems empty. > Is this mandatory? No, albeit recommended. Otherwise it reverts to the original pool placement system (probably what happened). > If so do i need to create the pools first? It's always recommended to create the pools first, that way you control the amount of pgs you want per pool. However, for such a test environment it shouldn't be an issue. > On Both cluster? Each zone is separate. > Should they be the same? Doesn't really matter, unless you're running everything on the same cluster, in which case each zone should have its own set of pools. Yehuda -- 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
