This commit introduces the necessary changes to support the new
messenger feature coming with Ceph Nautilus (currently in development).
What changes? Now the monitor listens on two port:
* old 6789 for messenger v1, which will help us support older client
(e,g: krbd)
* new 3300 for messengers v2, which brings new improvement in the
messaging layer. This new transport layer brings numerous advantages
such as encryption improvement, speed improvement, pluggable nature to
support different network stack than TCP and many more.
We still have one Service IP, however it has 2 ports, see:
```
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
rook-ceph-mon-a ClusterIP 10.106.217.160 <none> 3300/TCP,6789/TCP 4h
rook-ceph-mon-b ClusterIP 10.99.36.175 <none> 3300/TCP,6789/TCP 4h
rook-ceph-mon-c ClusterIP 10.108.220.74 <none> 3300/TCP,6789/TCP 4h
````
The `ceph.conf` has changed and we don't force the port when using an IP
address (public addr etc). Ceph, depending on its version will naturally
start the monitors on their right port, 6789.
A new --ceph-version-name CLI argument has been added to the Rook binary
so that when the pod starts it passes the ceph version name and the
configuration of the ceph.conf, as well as daemon startup flags, happen
properly.
Given that the Rook Operator remembers the port of all the monitors it
deployed (through Pod definition), this change is not an issue and will
maintain backward compatibility.
Note that to test this you must build rook with dev container image,
which contains the dev Nautilus version. So you should do something
like:
`make -j4 BASEIMAGE='ceph/daemon-base:latest-master' IMAGES='ceph' build`
Resolves: #2525
Signed-off-by: Sébastien Han <seb@redhat.com>
Configure the Ceph mgr daemon completely from the operator a la the
recent changes to the Ceph mon operator.
The pod spec for the mgr changed quite a bit, and instead of updating
the unit tests of questionable use, some additional unit test tools
applicable to any Ceph daemon have been added and used with the mgr. The
mgrs unit tests should now be more useful and get in the way of devs
less, and they can be used by other daemons later.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Make the Rook config-init unnecessary for mons, and remove that init
container. Perform all mon configuration steps in the operator, and set
up the mon pods and k8s environment such that only Ceph containers are
needed for running mons.
This should help streamline changes to the mons, as there will be no
need to change the `daemon/mon` code or `cmd/rook/ceph` code with mon
changes in the future.
This work starts to lay the groundwork for supporting the
`design/ceph-config-updates.md` design.
Notable new bits:
Create a keyring secret store helper for storing dameon keyrings, and
use it to store the mon keyring. Mon pods mount the keyring into a
k8s secret-backed volume.
Create a configmap store for the Ceph config file which can be mounted
into pods/containers directly to /etc/ceph/ceph.conf. Also store
individual mon_host and mon_initial_members values which can be mapped
into pods as environment variables and used in Ceph commandline flags,
enabling the mon pods to have the most up-to-date information about the
mon cluster when restarting and without need for operator intervention.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Set each Ceph mon's `public bind addr` from the commandline instead of
by inserting it into the config file. This eliminates an extra step from
initializing mon configs.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Rook's behavior should not be to create a default OSD in dataDirHostPath
when no devices are present on a node. Any preexisting fallback osd
should be kept as long as no dirs or disks have been specified to keep
the legacy behavior for those clusters running with these osds in place.
The legacy deletion behavior is also kept, removing the osd as soon as
any dir or device is specified other than the dataDirHostPath dir.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
The Ceph CLI has the flag option '--connect-timeout' which controls the
timeout of the command when reaching out to a monitor. If the connection
cannot be established within 15 seconds, the command will fail. The
default value in Ceph is relatively long (300sec IIRC). This is a
problem when monitors have trouble reaching out quorum, since this means
the command will only fail after 300 sec.
Now we fail faster.
Resolves: https://github.com/rook/rook/issues/2574
Signed-off-by: Sébastien Han <seb@redhat.com>
Introduce mount security mode for basic multi tenancy
Fixes#2164.
This adds three new parameters/options to StorageClass/flexvolume entry:
* `mountUser`
* `mountSecret`
* `mountSecretNamespace`
Signed-off-by: Alexander Trost <galexrt@googlemail.com>
This seems a bit misplaced, since CreateFilesystem is all about
doing the initial setup of the filesystem, and the daemons are
managed in a different block of code.
In a later patch, we're going to want to allow the CRD to spin
up MDS' for filesystems that already exist. Lift the code that
sets the MDS count into the caller.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Fixes#2239
Set Ceph logging to `/dev/stderr`, and run rgw setup commands without
combined output so that logging (no on stderr) isn't captured as part of
the result of commands.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Previous code tried to interpret the string output as a `fmt.Sprintf`
string formatter, mangling `%` into `%!(MISSING)`, etc.
Thanks to go, I don't belive these are exploitable (unlike the similar
error in C).
Since this seemed to be a common error in the codebase, I did a quick
audit by visually inspecting the results of `git grep 'f([^"]'`. I
don't have a good suggestion for automated tests to prevent this in
future :(
Example error (look for `(MISSING)`):
```
E0927 05:31:07.618429 11227 driver-call.go:237] Failed to unmarshal output for command: unmount, output: "2018-09-27 05:31:07.191711 I | exec: Running command: df --type ceph /var/lib/kubelet/pods/95461479-c216-11e8-bcf0-02030782ac80/volumes/ceph.rook.io~rook/oe-scratch\n2018-09-27 05:46:43.808596 I | Filesystem 1K-blocks Used Available Use%!M(MISSING)ounted on\n2018-09-27 05:46:43.808659 I | 10.107.25.147:6790,10.109.173.79:6790,10.104.85.255:6790:/ 151678976 49410048 102268928 33%!/(MISSING)var/lib/kubelet/pods/95461479-c216-11e8-bcf0-02030782ac80/volumes/ceph.rook.io~rook/oe-scratch\n{\"status\":\"Success\"}\n", error: invalid character '-' after top-level value
```
Signed-off-by: Angus Lees <gus@inodes.org>
fix#2193
Fix regression introduced by PR allowing MDS config creation in init
container. This allows the new deployment-based mds clusters to scale
down where currently they can only remain constant or scale up.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Progress toward issue #2003.
Includes design from design doc PR #1578
Use init containers to create configuration for Ceph mgrs. There is only
one init container in this design. The init container calls the Rook
binary to create Ceph config files which are then shared with the mds
daemon main container.
Once this init is run, the main mds daemon is run. Leaving room to use
the Ceph-versioned image in the future, call `ceph-mds --foreground ...`
to run the Ceph mds.
The refactor to using an init container also necessitated refactoring
the mdses replicaset implementation to a deployment-per-pod
implementation due to a chicken-egg problem. With a single container (in
the before times) the Rook binary was able to call the ceph-mds daemon
with an id generated from the pod name. Since the pod name is not known
before runtime, and the id is one of the few params that must be
specified to Ceph daemons on run, it is necessary to know the id
beforehand; thus the move to a deployment architecture following the
likes of the mon and mgr daemons.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Progress toward issue #2003.
Includes design from design doc PR #1578
Use init containers to create configuration for Ceph mgrs. There is only
1 init container in this design:
1. Using the Rook image, call the Rook binary to create Ceph config
files shared with the mgr daemgr container.
Once this init is run, the main mgr daemgr is run. Leaving room to use
the Ceph-versioned image in the future, call `ceph-mgr --foreground ...`
to run the Ceph mgr.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Add functions for helping test Kubernetes volumes and volume mounts.
- Create functions for testing the existence of volumes/mounts by
name in a list of vols/mounts without needing to know the index of
the vol/mount in the list.
- Create functions for printing vols/mounts in a human-readable format
so that tests may output more useful errors.
- Create a test definition for ensuring that all the volumes a pod
provides and all the mounts from each pod container match. For each
volume, there must be at least one container which mounts the volume,
and for each mount, there must be a source volume.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Progress toward issue #2003.
Includes design from design doc PR #1578
Use init containers to create configuration for Ceph mons. There are 3
init containers in this design:
1. Using the Rook image, call the Rook binary to create Ceph config
files ultimately stored in a k8s EmptyDir which is shared between all
containers in a mon pod.
2. Leaving room to use the Ceph-versioned image in the future, call
`monmaptool` to generate the monmap, stored in a shared config dir.
3. Leaving room to use the Ceph-versioned image in the future, call
`ceph-mon --mkfs ...` to set up the mon filesystem on the host's disk.
Once this init is run, the main mon daemon is run. Leaving room to use
the Ceph-versioned image in the future, call `ceph-mon --foreground ...`
to run the Ceph mon.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
While creating a PVC, if the operator dies between successfully creating a rbd
image and associating it with a PV, that PVC will never be completed. A
restarted operator will also fail to complete the task as the requested rbd
image already exists. With the assumption that a PVC UID will always be unique,
this patch will try to use the existing volume and complete the PVC request.
Resolves: #2072
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
Sébastien Han
Blaine Gardner
Travis Nielsen
Alexander Trost
Andrey Laguta
Bogdan Luca
Angus Lees
Ganesh Maharaj Mahalingam