Hello All,
I've made some progress, but I'm still having a bit of difficulty.
I've got all my monitors responding now, I had to break my openvswitch
configuration as ceph was trying to start, but since ovs doesn't start
networking until the os is up, the monitors were unable to communicate with
each other and were failing to start. Assigning the cluster IP to the
physical nic resolved this issue mostly. Except, kitt is now listening on
port 6800 instead of 6789 like all the other nodes:
root@kitt:~# ceph -s
health HEALTH_OK
monmap e8: 3 mons at {kitt=
192.168.0.35:6800/0,red6=192.168.0.40:6789/0,shepard=192.168.0.2:6789/0},
election epoch 136, quorum 0,1,2 kitt,red6,shepard
osdmap e939: 4 osds: 4 up, 4 in
pgmap v69300: 224 pgs: 224 active+clean; 571 GB data, 1143 GB used,
1648 GB / 2792 GB avail
mdsmap e1: 0/0/1 up
root@kitt:~# netstat -anp | grep ceph
tcp 0 0 192.168.0.35:6800 0.0.0.0:* LISTEN
1181/ceph-mon
tcp 0 0 192.168.0.35:58259 192.168.0.2:6789
ESTABLISHED 1181/ceph-mon
tcp 0 0 192.168.0.35:58324 192.168.0.40:6789
ESTABLISHED 1181/ceph-mon
unix 2 [ ACC ] STREAM LISTENING 12620 1181/ceph-mon
/var/run/ceph/ceph-mon.kitt.asok
This seems to be OK usually, however, intermittently, I'm getting an error
that the daemon cannot be contacted on 6789:
2013-06-17 11:30:10.773333 7fd1170b9700 0 -- :/29711 >>
192.168.0.35:6789/0pipe(0x236f510 sd=3 :0 s=1 pgs=0 cs=0 l=1).fault
How do I tell ceph to either run on port 6789 or tell it that this
particular monitor is running on 6800?
That said, I am unable to deploy OSDs on the kitt host. There were three
running OSDs before all this trouble, however, when my cluster crashed,
they would not come back up.
root@kitt:~# ceph osd tree
# id weight type name up/down reweight
-1 2.72 root default
-2 0.91 host red6
0 0.91 osd.0 up 1
-3 0 host kitt
-4 1.81 host shepard
3 0.45 osd.3 up 1
4 0.45 osd.4 up 1
6 0.91 osd.6 up 1
What had previously worked was removing the osd completely, then readding
it, however, when I attempt to readd the OSD, the journal is created, but
the mount point is not and the disk is never mounted:
root@shepard:~/ceph-conf# ceph-deploy osd --zap-disk create
kitt:/dev/sdb:/var/lib/ceph/osd/ceph-sdb.journal
root@kitt:~# ls -lah /var/lib/ceph/osd/
total 8.0K
drwxr-xr-x 2 root root 4.0K Jun 16 23:13 .
drwxr-xr-x 8 root root 4.0K Jun 1 23:23 ..
-rw-r--r-- 1 root root 1.0G Jun 12 20:14 ceph-osd-journal.sdb
-rw-r--r-- 1 root root 1.0G Jun 16 23:13 ceph-sdb.journal
The last recorded error was an "Unable to open superblock" error, that was
several weeks ago, but I think that coincides with the initial trouble I
experienced. I have tested these disks and can confirm that they have not
failed. I am able to format and mount them using either a GPT or MSDOSPT
and can read/write to them. I also ran the Smartmontools long test and a
full read/write test with another tool... (the name escapes me... badpart
or similar), so I don't think the actual disk is at fault here. (I
actually might have another disk that I can test with, but ideally I would
like to use the disks that were previously in use in this host).
Relevant logs:
root@kitt:~# tail /var/log/ceph/ceph-osd.1.log.5
-2/-2 (syslog threshold)
-1/-1 (stderr threshold)
max_recent 10000
max_new 1000
log_file /var/log/ceph/ceph-osd.1.log
--- end dump of recent events ---
2013-06-12 17:59:27.811593 7fb1d15817c0 0 ceph version 0.61.3
(92b1e398576d55df8e5888dd1a9545ed3fd99532), process ceph-osd, pid 5400
2013-06-12 17:59:27.811620 7fb1d15817c0 -1 ** ERROR: unable to open OSD
superblock on /var/lib/ceph/osd/ceph-1: (5) Input/output error
2013-06-12 18:57:27.689176 7f245f9957c0 0 ceph version 0.61.3
(92b1e398576d55df8e5888dd1a9545ed3fd99532), process ceph-osd, pid 4953
2013-06-12 18:57:27.702866 7f245f9957c0 -1 ** ERROR: unable to open OSD
superblock on /var/lib/ceph/osd/ceph-1: (2) No such file or directory
==> /var/log/ceph/ceph-osd.5.log.5 <==
2013-06-12 18:20:04.369439 7f984a5e77c0 -1 ** ERROR: error converting
store /var/lib/ceph/osd/ceph-5: (1) Operation not permitted
2013-06-12 18:24:20.252341 7f88625bc7c0 0 ceph version 0.61.3
(92b1e398576d55df8e5888dd1a9545ed3fd99532), process ceph-osd, pid 4955
2013-06-12 18:24:20.394257 7f88625bc7c0 0
filestore(/var/lib/ceph/osd/ceph-5) mount FIEMAP ioctl is supported and
appears to work
2013-06-12 18:24:20.394284 7f88625bc7c0 0
filestore(/var/lib/ceph/osd/ceph-5) mount FIEMAP ioctl is disabled via
'filestore fiemap' config option
2013-06-12 18:24:20.394829 7f88625bc7c0 0
filestore(/var/lib/ceph/osd/ceph-5) mount did NOT detect btrfs
2013-06-12 18:24:20.402567 7f88625bc7c0 0
filestore(/var/lib/ceph/osd/ceph-5) mount syncfs(2) syscall fully supported
(by glibc and kernel)
2013-06-12 18:24:20.402781 7f88625bc7c0 0
filestore(/var/lib/ceph/osd/ceph-5) mount found snaps <>
2013-06-12 18:24:20.457881 7f88625bc7c0 -1
filestore(/var/lib/ceph/osd/ceph-5) Error initializing leveldb: Corruption:
checksum mismatch
2013-06-12 18:24:20.457944 7f88625bc7c0 -1 ** ERROR: error converting
store /var/lib/ceph/osd/ceph-5: (1) Operation not permitted
Any help is greatly appreciated as I am really stumped.
I think my biggest frustration is the init scripts not working as described
in the docs. After I use ceph-deploy, do I need to write a config file?
based on my interpretation of the docs and upstart scripts, I don't think
so; the respective daemons start on boot...
Thanks for your time,
Jon A
---------- message ----------
From: Jon <three1...@gmail.com>
Date: Sun, Jun 9, 2013 at 12:36 PM
Subject: [ceph-users] Help Recovering Ceph cluster
To: ceph-users <ceph-us...@ceph.com>
Hello All,
I have deployed Ceph with ceph-deploy on three nodes. Each node is
currently play double duty as both a monitor node and an osd node.
This is what my cluster looked like before I rebooted the one node:
>> root@red6:~# ceph -s
>> health HEALTH_OK
>> monmap e2: 3 mons at {kitt=
192.168.0.35:6789/0,red6=192.168.0.40:6789/0,shepard=192.168.0.2:6789/0},
election epoch 10, quorum 0,1,2 kitt,red6,shepard
>> osdmap e29: 7 osds: 7 up, 7 in
>> pgmap v1692: 192 pgs: 192 active+clean; 19935 MB data, 40441 MB
used, 4539 GB / 4580 GB avail; 73B/s rd, 0op/s
>> mdsmap e1: 0/0/1 up
Last week, I rebooted one of my nodes (kitt), and after it rebooted, I was
unable to get it to rejoin the cluster.
Typically, I've been able to use ceph-deploy to delete the monitor then
deploy the monitor and everything goes back to normal.
That didn't work this time, so that monitor and those two osds were out of
the cluster. I think when I rebooted the node,
I got errors that the mon was unreachable and it was taken out of the
monmap. When I tried to delete the monitor ceph-deploy
was telling me there was not monitor to delete. When I ran ceph-deploy to
deploy the monitor I received no errors, but the monitor never started.
I've tried a number of things in the docs, but something seems amiss
because when I try to restart monitors or osds, the init script tells me
it's not found.
I've copied my ceph.conf at the end of this e-mail.
>> root@shepard:~# ls /var/lib/ceph/mon/
>> ceph-shepard
>> root@shepard:~# /etc/init.d/ceph restart mon.shepard
>> /etc/init.d/ceph: mon.shepard not found (/etc/ceph/ceph.conf defines ,
/var/lib/ceph defines )
(I've also tried mon.0 .. mon.3 )
Same thing with osds:
>> root@shepard:~# ls /var/lib/ceph/osd/
>> ceph-3 ceph-4 ceph-6 sdb.journal sdc.journal sdd.journal
>> root@shepard:~# /etc/init.d/ceph restart osd.4
>> /etc/init.d/ceph: osd.4 not found (/etc/ceph/ceph.conf defines ,
/var/lib/ceph defines )
(I've tried every iteration I can think of, using `service ceph restart`,
I've also tried both iterations with the -a flag e.g. `service ceph -a
restart`, and stop/start instead of restart, also with and without the -a
flag)
That's about where I left it last week, before my work week started. Other
than occasionally reporting something to the effect of:
>> 2013-06-09 11:46:14.978350 7f8098e9a700 0 -- :/18915 >>
192.168.0.35:6789/0 pipe(0x7f8090002ca0 sd=4 :0 s=1 pgs=0 cs=0 l=1).fault
when I ran ceph -s , ceph health reported OK, so I figured it was ok to
leave until I was off shift and had the cycles to devote to it...
I really do know better.
Last night I experienced a power outage, and all of my nodes hard rebooted.
Now when I run ceph health (or use any of the ceph or ceph-deploy tools),
I get the following error message:
>> 2013-06-09 12:12:14.692484 7f4506e0e700 0 -- :/24787 >>
192.168.0.40:6789/0 pipe(0x7f44fc000c00 sd=4 :0 s=1 pgs=0 cs=0 l=1).fault
>> 2013-06-09 12:12:17.692509 7f450d5a0700 0 -- :/24787 >>
192.168.0.35:6789/0 pipe(0x7f44fc003010 sd=4 :0 s=1 pgs=0 cs=0 l=1).fault
>> 2013-06-09 12:12:20.692469 7f4506e0e700 0 -- :/24787 >>
192.168.0.2:6789/0 pipe(0x7f44fc0038c0 sd=4 :0 s=1 pgs=0 cs=0 l=1).fault
this was run on the shepard node (192.168.0.2), but I get the same error
from all of the nodes.
I do see a number of ceph processes
>> root@shepard:~# ps aux | grep ceph
>> root 1041 0.0 0.0 33680 7100 ? Ss 06:11 0:18
/usr/bin/python /usr/sbin/ceph-create-keys --cluster=ceph -i shepard
>> root 1538 0.0 0.0 378904 4284 ? Sl 06:11 0:04 ceph
--cluster=ceph --name=osd.3 --keyring=/var/lib/ceph/osd/ceph-3/keyring osd
crush create-or-move -- 3 0.45 root=default host=shepard
>> root 1581 0.0 0.0 378904 4300 ? Sl 06:11 0:04 ceph
--cluster=ceph --name=osd.6 --keyring=/var/lib/ceph/osd/ceph-6/keyring osd
crush create-or-move -- 6 0.91 root=default host=shepard
>> root 1628 0.0 0.0 378904 4296 ? Sl 06:11 0:04 ceph
--cluster=ceph --name=osd.4 --keyring=/var/lib/ceph/osd/ceph-4/keyring osd
crush create-or-move -- 4 0.45 root=default host=shepard
However, I don't see the monitor process actually running.
My first question is, how can I restart ceph processes?
My second question is, am I on the right track here? My hypothesis is
starting the monitors will get things going again, because there is nothing
is listening on those ports:
>> root@shepard:~# telnet 192.168.0.2 6789
>> Trying 192.168.0.2...
>> telnet: Unable to connect to remote host: Connection refused
>> root@shepard:~# netstat -anp | grep ceph
>> root@shepard:~# netstat -anp | grep 6789
Actually, I just rebooted this node again and now the osds are not running.
only ceph-create-keys.
My last question is where does ceph-deploy create the configs? I have the
original files in a directory where I ran ceph deploy,
and I know about the /etc/ceph/ceph.conf file, but there seems to be some
other config that the cluster is pulling from.
Maybe I'm mistaken, but there are no osds in my ceph.conf.
Thanks for all your help.
Jon A
ceph.conf file:
[global]
fsid = 86f293d0-bec7-4694-ace7-af3fbaa98736
mon_initial_members = red6, kitt, shepard
mon_host = 192.168.0.40,192.168.0.35,192.168.0.2
auth_supported = cephx
osd_journal_size = 1024
filestore_xattr_use_omap = true
_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
