Bug #9602
closedrbd export -> nc ->rbd import = memory leak
0%
Description
I see a memory leak when importing raw devi?e.
Export Scheme:
[rbd@rbdbackup ~]$ rbd --no-progress -n client.rbdbackup -k /etc/ceph/big.keyring -c /etc/ceph/big.conf export rbdtest/vm-111-disk-1 - | nc 10.43.255.252 12345
[root@ct2 ~]# nc -l 12345 | rbd import --no-progress --image-format 2 - rbd/vm-111-disk-1
This is the same problem with ssh.
Memory usage, see the screenshots.
OS: CentOS 6.5
Kernel: 2.6.32-431.el6.x86_64
Ceph --version: ceph version 0.72.2 (a913ded2ff138aefb8cb84d347d72164099cfd60)
Files
Updated by Sage Weil over 9 years ago
- Project changed from Ceph to rbd
- Category deleted (
librbd) - Priority changed from Normal to High
Updated by Irek Fasikhov over 9 years ago
Hi, Sage.
I'm sorry, was wrong to put up parameter: rbd_cache size
The problem is not confirmed.
Updated by Irek Fasikhov over 9 years ago
As a suggestion, prohibit the use of the cache when RDB imports.
Updated by Jason Dillaman over 9 years ago
I quickly attempted to reproduce this on the same version w/o success. Can you attach /etc/ceph/big.conf? How large is the image that you are importing?
Updated by Josh Durgin over 9 years ago
- Status changed from New to Need More Info
Updated by Irek Fasikhov over 9 years ago
Jason Dillaman wrote:
I quickly attempted to reproduce this on the same version w/o success. Can you attach /etc/ceph/big.conf? How large is the image that you are importing?
Hi, Jason.
It was not a mistake in the code, I made ??a mistake with the parameter rbd_cache_size:
rbd_cache size = 134217728000 //130 GB :)
Updated by Jason Dillaman over 9 years ago
- Status changed from Need More Info to Closed
Irek, thanks for the update. Closing as not a bug.