Hello all,
I am trying to check out how Ceph is doing under Scientific Linux 5.5
(unfortunately, academia loves SL).
Currently I am running a cluster of 2 VMs, VM 1 is mon, mds and osd, VM 2 is
running osd only, and VM 3 is where I mount ceph. All VMs are running SL5.5
x86_64 with the latest updates and are on a gbit network. Additionally I
compiled ceph from source, (checked out yesterday), with fuse support, since
SL5.5 has a 2.6.18 kernel...
Using the fuse client, and a very simple configuration file, I am facing the
following issues:
- When adding files 120mb< to the filesystem, the following messages appear,
and the the writing speed slows down (mon reports that all authenticated osds
are up and running)
client4314.objecter pg 0.aae5 on [1,0] is laggy: 157
client4314.objecter pg 0.d2d0 on [0,1] is laggy: 175
client4314.objecter pg 0.b7a6 on [1,0] is laggy: 171
client4314.objecter pg 0.6031 on [0,1] is laggy: 173
- Additionally, when trying to delete a file, cfuse crashes:
./include/xlist.h: In function 'xlist<T>::~xlist() [with T =
ObjectCacher::Object*]':
./include/xlist.h:63: FAILED assert(_size == 0)
1: (xlist<ObjectCacher::Object*>::~xlist()+0x30) [0x658ed4]
2: (ObjectCacher::ObjectSet::~ObjectSet()+0x2f) [0x658f51]
3: (Inode::~Inode()+0x10e) [0x673b68]
4: (Client::put_inode(Inode*, int)+0x10f) [0x6181d3]
5: (Client::_ll_put(Inode*, int)+0x116) [0x61866a]
6: (Client::ll_forget(vinodeno_t, int)+0x37a) [0x618a06]
7: cfuse [0x605db1]
8: (fuse_session_loop()+0x75) [0x2af02b790de5]
9: (ceph_fuse_ll_main(Client*, int, char const**)+0x24a) [0x60431a]
10: (main()+0x275) [0x5e1835]
11: (__libc_start_main()+0xf4) [0x3d7361d994]
12: (__gxx_personality_v0()+0x2a1) [0x5e1479]
NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to
interpret this.
./include/xlist.h: In function 'xlist<T>::~xlist() [with T =
ObjectCacher::Object*]':
./include/xlist.h:63: FAILED assert(_size == 0)
1: (xlist<ObjectCacher::Object*>::~xlist()+0x30) [0x658ed4]
2: (ObjectCacher::ObjectSet::~ObjectSet()+0x2f) [0x658f51]
3: (Inode::~Inode()+0x10e) [0x673b68]
4: (Client::put_inode(Inode*, int)+0x10f) [0x6181d3]
5: (Client::_ll_put(Inode*, int)+0x116) [0x61866a]
6: (Client::ll_forget(vinodeno_t, int)+0x37a) [0x618a06]
7: cfuse [0x605db1]
8: (fuse_session_loop()+0x75) [0x2af02b790de5]
9: (ceph_fuse_ll_main(Client*, int, char const**)+0x24a) [0x60431a]
10: (main()+0x275) [0x5e1835]
11: (__libc_start_main()+0xf4) [0x3d7361d994]
12: (__gxx_personality_v0()+0x2a1) [0x5e1479]
NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to
interpret this.
terminate called after throwing an instance of 'ceph::FailedAssertion*'
Aborted
---
Ceph.conf
[global]
auth supported = cephx
keyring = /etc/ceph/keyring.bin
[mon]
mon data = /opt/pragmata/monitor0
debug ms = 1
[mon0]
host = mpriki1
mon addr = 10.0.0.10:6789
[mds]
keyring = /etc/ceph/keyring.$name
[mds.mpriki1]
host = mpriki1
[osd]
osd data = /opt/pragmata/data/osd$id
osd journal = /opt/pragmata/data/osd$id/journal
keyring = /opt/ceph/keyring.$name
debug osd = 1
debug filestore = 1
[osd0]
host = mpriki1
osd journal size = 100
[osd1]
host = mpriki2
osd journal size = 100
----------
Under the same setup and ceph v0.20.2, client doesn't crash when I delete a
file, the file looks deleted, but even after a while, ceph has not freed any
space. The problem with the laggy client, is present in v0.20.2 as well.
I reckon that installing a newer kernel is one solution, and getting rid of
SL5.5 once and for all is possibly another one, but any ideas on what's
causing those issues in this setup, are welcomed.
Cheers,
-effie--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
