CEPH FS Storage Driver

olivierlambert

Hi @Rainer and welcome

Please use Markdown syntax for console/code blocks:

```
my test
```

Rainer

Hello oliverlambert,
thanks for the hint. I edited my post correcting the formatting.

r1

@Rainer Please share your NFSSR.py to check. Are you able to manually mount the SR using #mount.ceph addr1,addr2,addr3,addr4:remotepath localpath?

Rainer

Hello,

thanks for your answer. I can mount the the CephFS filesystem on the xenserver this way:

mount.ceph  1.2.3.4:/base /mnt -o "name=admin,secretfile=/etc/ceph/admin.secret"

where 1.2.3.4 is the IP of the active ceph monitor.

In xencenters New-SR-dialog I put the "1.2.3.4:/base" part in the "Share Name" input field and the options name=admin,secretfile=/etc/ceph/admin.secret in the "Advanced Options" field .

Below I attached my patched /opt/xensource/sm/NFSSR.py file:
[0_1569574207746_NFSSR.py](Uploading 100%)

NFSSR.py.txt

r1

Just so you know - the original patch was for XCP-NG 7.6. I'm checking it for XCP-NG 8.

r1

Can you share /opt/xensource/sm/nfs.py as well... it was also supposed to be patched.

r1

Updated new patch for XCP-NG 8.

olivierlambert

Note that all are hacks are won't survive any upgrade and aren't supported But it's fine for a proof of concept.

r1

Agree - unless we have enough interest from community to pack it or make it maintainable, it will stay a hack.

Also mentioning the command to apply patch # patch -d/ -p1 < cephfs-8.patch

Edit : updated patch strip level

Rainer

Hello and thanks for the new patch. I understand that this is only a kind of hack or better a proof of concept. I do not want to use this for a productive system but only wanted to give it a try with nautilus and see what performance I can expect if this hack hopefully develops to a supported integration of Ceph.

I hoped that the patch might work for XCP-ng 8.0 as well but it seems it does not that easily. So I could also install a xenserver 7.6 and use the original patch and test how it works for me. I do not want you to invest too much work in the 8.0 adoption. So if you think testing the patch 8.0 does not really help you in any way, then please do not hesitate and tell me. Else if you are as well interested to test if it works with 8.0 here is what I found out:

The new patch does not work completely, there is a "patch unexpectedly ends in middle of line" -warning. On a fresh installed system:

# cd /
# patch --verbose -p0  < /cephfs-8.patch
mm...  Looks like a unified diff to me...
The text leading up to this was:
--------------------------
|--- /opt/xensource/sm/nfs.py   20199-27 16:45:11.918853933 +0530
|+++ /opt/xensource/sm/nfs.py   20199-27 16:49:26.910379515 +0530
--------------------------
patching file /opt/xensource/sm/nfs.py
Using Plan A...
Hunk #1 succeeded at 137.
Hunk #2 succeeded at 159.
Hmm...  The next patch looks like a unified diff to me...
The text leading up to this was:
--------------------------
|--- /opt/xensource/sm/NFSSR.py 20199-27 16:44:56.437557255 +0530
|+++ /opt/xensource/sm/NFSSR.py 20199-27 17:07:45.651188670 +0530
--------------------------
patching file /opt/xensource/sm/NFSSR.py
Using Plan A...
Hunk #1 succeeded at 137.
Hunk #2 succeeded at 160.
patch unexpectedly ends in middle of line
Hunk #3 succeeded at 197 with fuzz 1.
done

I can do a ceph.mount just like described before but when I instead try to create a new Ceph-FS-SR named "Ceph" in xencenter 8.0.0 I still get the error that SM has thrown a generic python exception.

The SMlog error log indicates that creating a NFS repos is tried with the given parameters and not a CephFS one:

...
Sep 30 13:03:05 rzinstal4 SM: [975] sr_create {'sr_uuid': '79cc55db-ee9f-9cc0-8d94-8cdb2915363b', 'subtask_of': 'DummyRef:|6456510f-b2dd-4dcd-b6ab-eb5eb40ceb9c|SR.create', 'args': ['0'], 'host_ref': 'OpaqueRef:72898e54-20b2-4046-9825-94553457fa60', 'session_ref': 'OpaqueRef:8a0819af-8b91-4086-a9cf-04ae37dc8e79', 'device_config': {'server': '1.2.3.4', 'SRmaster': 'true', 'serverpath': '/base', 'options': 'name=admin,secretfile=/etc/ceph/admin.secret'}, 'command': 'sr_create', 'sr_ref': 'OpaqueRef:91c592f6-47a0-40d6-8a33-7fe01831145f'}
Sep 30 13:03:05 rzinstal4 SM: [975] ['/usr/sbin/rpcinfo', '-p', '1.2.3.4']
Sep 30 13:03:05 rzinstal4 SM: [975] FAILED in util.pread: (rc 1) stdout: '', stderr: 'rpcinfo: can't contact portmapper: RPC: Remote system error - Connection refused
Sep 30 13:03:05 rzinstal4 SM: [975] '
Sep 30 13:03:05 rzinstal4 SM: [975] Unable to obtain list of valid nfs versions
...

I attached my /opt/xensource/sm/nfs.py file as well.

Thanks a lot

Rainer
nfs.py.txt

r1

@r1 said in CEPH FS Storage Driver:

patch -d/ -p0 < cephfs-8.patch

Please try # patch -d/ -p0 < cephfs-8.patch

r1

I see that there is a call to check_server_tcp which is expected to return True but not doing so in my patch. I'll modify it.

r1

The same patch is updated. Kindly download again and try with # patch -d/ -p1 < cephfs-8.patch.

Rainer

Thanks for the new patch. Applying the patch does not report any errors or warnings. When I try to create a "NFS" SR I still get the generic python error message from xencenter however the error message in /var/log/SMlog has changed:

Oct  1 10:17:52 xenserver8 SM: [4975] lock: opening lock file /var/lock/sm/76ad6fee-a7a8-34e6-bcce-a53f69a677a9/sr
Oct  1 10:17:52 xenserver8 SM: [4975] lock: acquired /var/lock/sm/76ad6fee-a7a8-34e6-bcce-a53f69a677a9/sr
Oct  1 10:17:52 xenserver8 SM: [4975] sr_create {'sr_uuid': '76ad6fee-a7a8-34e6-bcce-a53f69a677a9', 'subtask_of': 'DummyRef:|c98cee36-24ff-4575-b7
43-3a84cd93ed17|SR.create', 'args': ['0'], 'host_ref': 'OpaqueRef:72898e54-20b2-4046-9825-94553457fa60', 'session_ref': 'OpaqueRef:175293cb-a791-
44be-bd5f-a4cb911adddb', 'device_config': {'server': '1.2.3.4', 'SRmaster': 'true', 'serverpath': '/base', 'options': 'name=admin,secretfil
e=/etc/ceph/admin.secret'}, 'command': 'sr_create', 'sr_ref': 'OpaqueRef:a36f0371-1dc7-4685-a208-24e04a1c1e08'}
Oct  1 10:17:52 xenserver8 SM: [4975] lock: released /var/lock/sm/76ad6fee-a7a8-34e6-bcce-a53f69a677a9/sr
Oct  1 10:17:52 xenserver8 SM: [4975] ***** generic exception: sr_create: EXCEPTION <type 'exceptions.NameError'>, global name 'useroptions' is no
t defined
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 110, in run
Oct  1 10:17:52 xenserver8 SM: [4975]     return self._run_locked(sr)
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 159, in _run_locked
Oct  1 10:17:52 xenserver8 SM: [4975]     rv = self._run(sr, target)
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 323, in _run
Oct  1 10:17:52 xenserver8 SM: [4975]     return sr.create(self.params['sr_uuid'], long(self.params['args'][0]))
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/NFSSR", line 216, in create
Oct  1 10:17:52 xenserver8 SM: [4975]     raise exn
Oct  1 10:17:52 xenserver8 SM: [4975]
Oct  1 10:17:52 xenserver8 SM: [4975] ***** NFS VHD: EXCEPTION <type 'exceptions.NameError'>, global name 'useroptions' is not defined
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 372, in run
Oct  1 10:17:52 xenserver8 SM: [4975]     return self._run_locked(sr)
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 159, in _run_locked
Oct  1 10:17:52 xenserver8 SM: [4975]     rv = self._run(sr, target)
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/SRCommand.py", line 323, in _run
Oct  1 10:17:52 xenserver8 SM: [4975]     return sr.create(self.params['sr_uuid'], long(self.params['args'][0]))
Oct  1 10:17:52 xenserver8 SM: [4975]   File "/opt/xensource/sm/NFSSR", line 216, in create
Oct  1 10:17:52 xenserver8 SM: [4975]     raise exn
Oct  1 10:17:52 xenserver8 SM: [4975]

r1

Looks like variable declaration issue. I had it working for XCP-NG 7.6... Give me some time to fix this and make it working for 8.0

jmccoy555

@r1 Is this still in development; would like to get it working with 8.1?

Manually mounting the CephFS works, but the patch fails, probably not surprisingly; or is a GlusterFS approach a better option?

With Ceph Octopus and the cephadm approach setting Ceph up is really easy, so maybe this will become a more popular feature....?

r1

@jmccoy555 I did not continue this. Will definitely rejuvenate it once 8.1 is released. Hopefully next week.

r1

@jmccoy555

Notice: This is experimental and will be replaced by any updates to sm package

Ok, so here is updated patch for XCP-NG 8.1

# cd /
# wget "https://gist.githubusercontent.com/rushikeshjadhav/ea8a6e15c3b5e7f6e61fe0cb873173d2/raw/85acc83ec95cba5db7674572b79c403106b48a48/ceph-8.1.patch"
# patch -p0 < ceph-8.1.patch

Apart from patch you would need to install ceph-common to obtain mount.ceph

# yum install centos-release-ceph-nautilus --enablerepo=extras
# yum install ceph-common

There is no need of ceph.ko as now CephFS is part of kernel in 8.1.

You can refer NFS SR (Ceph) creation screenshot in earlier post or use following
# xe sr-create type=nfs device-config:server=10.10.10.10 device-config:serverpath=/ device-config:options=name=admin,secretfile=/etc/ceph/admin.secret name-label=CephFS

To remove this and restore to normal state
# yum reinstall sm

Note: Keep secret in /etc/ceph/admin.secret with permission 600

Edit: Hope to get this into a proper CephSR someday. Until then it must be considered for Proof of Concept only.
Changed ceph release repo to nautilus.

jmccoy555

@r1 Hi, just given this a try, but with no luck....

Manual mount works and I can see files in /mnt/cephfs

mount -t ceph 10.10.1.141:/xcp /mnt/cephfs -o name=admin,secret=AQDZVn9eKD4hBBAAvf5GGWcRFQAZ1eqVPDynOQ==

Trying as per the instructions.....

xe sr-create type=nfs device-config:server=10.10.1.141 device-config:serverpath=/xcp device-config:options=name=admin,secretfile=/etc/ceph/admin.secret name-label=CephFS
Error code: SR_BACKEND_FAILURE_73
Error parameters: , NFS mount error [opterr=mount failed with return code 2],

And without the secret file just to check that wasn't the issue....

xe sr-create type=nfs device-config:server=10.10.1.141 device-config:serverpath=/xcp device-config:options=name=admin,secret=AQDZVn9eKD4hBBAAvf5GGWcRFQAZ1eqVPDynOQ== name-label=CephFS
Error code: SR_BACKEND_FAILURE_108
Error parameters: , Unable to detect an NFS service on this target.,

Last round in the log file

Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] lock: opening lock file /var/lock/sm/94fa22bb-e9a9-45b2-864a-392117b17891/sr
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] lock: acquired /var/lock/sm/94fa22bb-e9a9-45b2-864a-392117b17891/sr
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] sr_create {'sr_uuid': '94fa22bb-e9a9-45b2-864a-392117b17891', 'subtask_of': 'DummyRe f:|b72aac42-1790-40da-ba29-ed43a6554094|SR.create', 'args': ['0'], 'host_ref': 'OpaqueRef:c0ae3b47-b170-4d48-9397-3c973872a8b 2', 'session_ref': 'OpaqueRef:85070c69-888a-4827-8d19-ffd53944aa16', 'device_config': {'server': '10.10.1.141', 'SRmaster': ' true', 'serverpath': '/xcp', 'options': 'name=admin,secretfile=/etc/ceph/admin.secret'}, 'command': 'sr_create', 'sr_ref': 'O paqueRef:f775fe87-0d82-4029-a268-dd2cdf4feb27'}
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] _testHost: Testing host/port: 10.10.1.141,2049
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] _testHost: Connect failed after 2 seconds (10.10.1.141) - [Errno 111] Connection ref used
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] Raising exception [108, Unable to detect an NFS service on this target.]
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] lock: released /var/lock/sm/94fa22bb-e9a9-45b2-864a-392117b17891/sr
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] ***** generic exception: sr_create: EXCEPTION <class 'SR.SROSError'>, Unable to dete ct an NFS service on this target.
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 110, in run
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     return self._run_locked(sr)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 159, in _run_locked
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     rv = self._run(sr, target)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 323, in _run
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     return sr.create(self.params['sr_uuid'], long(self.params['args'][0]))
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/NFSSR", line 198, in create
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     util._testHost(self.dconf['server'], NFSPORT, 'NFSTarget')
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/util.py", line 915, in _testHost
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     raise xs_errors.XenError(errstring)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476] ***** NFS VHD: EXCEPTION <class 'SR.SROSError'>, Unable to detect an NFS service on  this target.
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 372, in run
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     ret = cmd.run(sr)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 110, in run
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     return self._run_locked(sr)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 159, in _run_locked
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     rv = self._run(sr, target)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/SRCommand.py", line 323, in _run
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     return sr.create(self.params['sr_uuid'], long(self.params['args'][0]))
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/NFSSR", line 198, in create
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     util._testHost(self.dconf['server'], NFSPORT, 'NFSTarget')
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]   File "/opt/xensource/sm/util.py", line 915, in _testHost
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]     raise xs_errors.XenError(errstring)
Apr  1 15:05:11 xcp-ng-bad-3 SM: [25476]

Is it still looking for a real NFS server?

edit: its logging port 2049 not 6789?

r1

@jmccoy555 please refer the note about the secret to be put in a file.