When I upload a new CDROM image through Sunstone to the CEPH datastore, I get the following error after some time:
[ImageAllocate] Error allocating a new image. NAME is already taken by IMAGE 74.
The image 74 is that newly created image, it is in LOCKED state, and it eventually switches to the READY state. The oned.log says the following about the image 74:
Tue Apr 26 09:17:51 2016 [Z0][ReM][D]: Req:2384 UID:0 ImageAllocate invoked , "NAME="Win 2008r2 Ser...", 100
Tue Apr 26 09:17:52 2016 [Z0][ImM][I]: Copying /var/tmp/3166720000-WindowsServerStdEntDCWeb2008R2withSP164-bit-ENiso to repository for image 74
Tue Apr 26 09:17:52 2016 [Z0][ReM][D]: Req:2384 UID:0 ImageAllocate result SUCCESS, 74
Tue Apr 26 09:17:52 2016 [Z0][ReM][D]: Req:8400 UID:0 ImageInfo invoked , 74
Tue Apr 26 09:17:52 2016 [Z0][ReM][D]: Req:8400 UID:0 ImageInfo result SUCCESS, "<IMAGE><ID>74</ID><U..."
Tue Apr 26 09:17:57 2016 [Z0][ReM][D]: Req:9232 UID:0 ImageInfo invoked , 74
Tue Apr 26 09:17:57 2016 [Z0][ReM][D]: Req:9232 UID:0 ImageInfo result SUCCESS, "<IMAGE><ID>74</ID><U..."
[the ImageInfo lines repeat every five seconds, I am omitting them]
Tue Apr 26 09:18:51 2016 [Z0][ReM][D]: Req:8768 UID:0 ImageAllocate invoked , "NAME="Win 2008r2 Ser...", 100
Tue Apr 26 09:18:51 2016 [Z0][ReM][E]: Req:8768 UID:0 ImageAllocate result FAILURE [ImageAllocate] Error allocating a new image. NAME is already taken by IMAGE 74.
[...]
Tue Apr 26 09:21:23 2016 [Z0][ReM][D]: Req:272 UID:0 ImageInfo invoked , 74
Tue Apr 26 09:21:23 2016 [Z0][ReM][D]: Req:272 UID:0 ImageInfo result SUCCESS, "<IMAGE><ID>74</ID><U..."
Tue Apr 26 09:21:26 2016 [Z0][ImM][I]: Image (74) copied and ready to use.
It seems that image creation (upload from /var/tmp/ to CEPH) takes longer than one minute, and oned decides to restart the action after one minute. This of course fails, because the image has already been created. The image in question has about 3 GB.
What is the best solution for this problem? Increasing the timeout somewhere? Making oned know that the upload is still in progress (as opposed to be dead), and that the operation does not need to be restarted yet?
Thanks!