Bug #3766
closedrbd resize command fails on a mixed node cluster when it is a copied rbd image and when same image is resized more than once
0%
Description
ubuntu@burnupi24:/var/log/ceph$ ceph -v
ceph version 0.56-131-gd283abd (d283abdf50b1e4429b775680bfae1bb20c75306b)
On a mixed node cluster, [3 nodes burnupi21 running osd.1, burnupi22 running osd.2, burnupi23 running osd.3 on argonaut and 1 node burnupi24 running osd.4 on v0.56],
There are two issues,
1. trying to resize a image that is a copied rbd image fails when the rbd image was created on the old osds running argonaut.
ubuntu@burnupi24:/var/log/ceph$ rbd create bob --size 100
ubuntu@burnupi24:/var/log/ceph$ ceph osd map rbd bob.rbd
osdmap e329 pool 'rbd' (2) object 'bob.rbd' -> pg 2.e589a57b (2.7b) -> up [3,2] acting [3,2]
ubuntu@burnupi24:/var/log/ceph$ rbd resize bob --size 1
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd cp bob bob1
Image copy: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize bob1 --size 10
2013-01-08 15:31:08.415096 7f95fd350780 -1 librbd: error registering a watch: (75) Value too large for defined data type
rbd: error opening image bob1: (75) Value too large for defined data type
ubuntu@burnupi24:/var/log/ceph$ ceph osd map rbd bob1.rbd
osdmap e329 pool 'rbd' (2) object 'bob1.rbd' -> pg 2.7d3bed1e (2.11e) -> up [2,3] acting [2,3]
2. trying to resize the same image more than once fails, if the rbd image is created on the old osds running argonaut.
ubuntu@burnupi24:/var/log/ceph$ rbd create bobnew --size 100
ubuntu@burnupi24:/var/log/ceph$ ceph osd map rbd bobnew.rbd
osdmap e329 pool 'rbd' (2) object 'bobnew.rbd' -> pg 2.b3c89d78 (2.78) -> up [2,1] acting [2,1]
ubuntu@burnupi24:/var/log/ceph$ rbd resize bobnew --size 11
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize bob1 --size 0
2013-01-08 15:38:55.120514 7fbcfac66780 -1 librbd: error registering a watch: (75) Value too large for defined data type
rbd: error opening image bob1: (75) Value too large for defined data type
ubuntu@burnupi24:/var/log/ceph$ rbd resize bobnew --size 9
2013-01-08 15:39:02.916251 7f5982b40780 -1 librbd: error registering a watch: (75) Value too large for defined data type
rbd: error opening image bobnew: (75) Value too large for defined data type
But it works fine, when the rbd image is created on the new osds running v0.56
ubuntu@burnupi24:/var/log/ceph$ rbd create onnew --size 100
ubuntu@burnupi24:/var/log/ceph$ ceph osd map rbd onnew.rbd
osdmap e329 pool 'rbd' (2) object 'onnew.rbd' -> pg 2.3f44bd84 (2.84) -> up [4,2] acting [4,2]
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew --size 90
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew --size 80
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew --size 70
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew --size 60
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew --size 50
Resizing image: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd cp onnew onnew1
Image copy: 100% complete...done.
ubuntu@burnupi24:/var/log/ceph$ rbd resize onnew1 --size 40
Resizing image: 100% complete...done.
Updated by Dan Mick over 11 years ago
- Status changed from New to Need More Info
I think this might be e1776809031c6dad441cfb2b9fac9612720b9083, which is still in next. Can you try an rbd client from master (or anything containing c4370ff03f8ab655a009cfd9ba3a0827d8c58b11, merged in last Friday)?
Updated by Tamilarasi muthamizhan over 11 years ago
This works fine on the master branch that has a fix for it :
ceph version 0.56-193-g00898c1 (00898c1860e8ae95b5219257d1635b15ccdce5c1)
Updated by Dan Mick over 11 years ago
- Status changed from Need More Info to Resolved
I'm calling this fixed, then.