This new windows provides more detailes about an OSD such as:
* PID
* Memory usage
* various metadata that could be of interest
* list of phyiscal disks used for the main disk, db and wal with
additional infos about the volumes for each
A new 'Details' button is added to the OSD overview and a double click
on an OSD will also open this new window.
The componend defines the items in the initComponent instead of
following a fully declarative approach. This is because we need to pass
the same store to multiple Proxmox.ObjectGrids.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Reviewed-by: Dominik Csapak <d.csapak@proxmox.com>
Tested-by: Dominik Csapak <d.csapak@proxmox.com>
Since the rule selector is not allowed to be empty, but the loading
of the rules is not instant, the validity change will trigger before
the load was finished. Since it is in the advanced section, it will
be opened every time instead of only when there is an invalid value.
This patch fixes that by temporarily setting 'allowBlank' to true
until the store is loaded, and then it revalidates the field.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Tested-By: Aaron Lauterer <a.lauterer@proxmox.com>
add support for setting the background and text color via css. also
allows for dynamically switching the color when a theme change is
detected.
Signed-off-by: Stefan Sterz <s.sterz@proxmox.com>
inline the transformation for the health store and also avoid setting
raw data from the outside
and drop some bogus comments along the way, first one should mostly
use "why?" not "what happens?" comments and second, commenting
straight forward things always makes one pause and recheck everything
far to often, as a comment indicates there something non-obvious
happening.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Some users have a more complicated CRUSH hierarchy, for example with a
stretched cluster. The additional hierarchy steps (datacenter, rack,
room, ...) are shown in the OSD panel. Showing a generic icon for any
CRUSH types that have not a specific icon configured will make it easier
to navigate the tree as it will not look somewhat broken and empty.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
By switching from 'ceph osd tree' to the 'ceph osd df tree' mon API
equivalent , we get the same data structure with more information per
OSD. One of them is the number of PGs stored on that OSD.
The number of PGs per OSD is an important number, for example when
trying to figure out why the performance is not as good as expected.
Therefore, adding it to the OSD overview visible by default should
reduce the number of times, one needs to access the CLI.
Comparing runtime cost on a 3 node ceph cluster with 4 OSDs each doing 50k
iterations gives:
Rate osd-df-tree osd-tree
osd-df-tree 9141/s -- -25%
osd-tree 12136/s 33% --
So, while definitively a bit slower, but it's still in the µs range,
and as such below HTTP in TLS in TCP connection setup for most users,
so worth the extra useful information.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
[ TL: slight rewording of subject and add benchmark data ]
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Check if stopping of a service (OSD, MON, MDS) will be problematic for
Ceph. The warning still allows the user to proceed.
Ceph also has a check if the destruction of a MON is okay, so let's use
it.
Instead of the common OK button, label it with `Stop OSD` and so forth
to hopefully reduce the "click OK by habit" incidents.
This will not catch it every time as Ceph can need a few moments after a
change to establish its current status. For example, stopping one of 3
MONs and then right away destroying one of the last two running MONs
will most likely not trigger the warning. Doing so after a few seconds
should show the warning though.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
If an OSD is removed during the wrong conditions, it could lead to
blocked IO or worst case data loss.
Check against global flags that limit the capabilities of Ceph to heal
itself (norebalance, norecover, noout) and if there are degraded
objects.
Unfortunately, the 'safe-to-destroy' Ceph API endpoint will not help
here as it only works as long as the OSD is still running. By the time
the destroy button is enabled, the OSD will already be stopped.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Note that we still check the cluster for an already used installation
and will select that, if any, so this is really just for setting up a
completely new cluster.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Removes the possibility to select the node on which to create the first
monitor in the configuration / initialization step and always sets it to
the current node.
This prevents that a user might select another node on which the Ceph
packages have not yet been installed. If a user did that, they would get
an error, but the Ceph config file would have been written. If the user
then does not select a valid node to create the first mon, but aborts
the wizard, they are greeted with a rados_connect error because the
config file exists, but it does not contain any mon infos that are
needed to connect to the Ceph cluster.
Creating a mon manually will remedy such a situation, but especially for
new users, this behavior is not ideal and confusing.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
By not auto filling the Ceph public network we can avoid accidental
clicks on 'Next' which will cause the first Mon to be created with a
potentially wrong network. While that is fixable, it is tedious and
can be easily avoided by making the user always select the network to
use.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Tested-by: Stefan Hrdlicka<s.hrdlicka@proxmox.com>
[ T: adapted commit subject to be more specific and match our common
style ]
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
They cannot be changed after pool creation for erasure coded pools
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Reviewed-by: Dominik Csapak <d.csapak@proxmox.com>
Tested-by: Dominik Csapak <d.csapak@proxmox.com>
The in & out commands for OSDs are not node specific and can be run on
any node in the Ceph cluster. By sending them to the node currently used
to access the UI they can still be sent even if the node on which the
OSDs are located is down.
This helps in a disaster scenario where a node is down. By default Ceph
will mark a downed OSD as out after 10 minutes. This could be too long
in some situations. Running the CLI command to mark the OSD as out
earlier on one of the remaining nodes does work, but if the admin is not
used doing it this way, this adds stress, in a potentially already
stressful situation.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Showing already configured custom device classes makes it easier to
create new OSDs with custom device classes.
The Crush map contains a list of all OSDs in the cluster, including
their device class.
This means we can create a list of used device classes from it, avoiding
adding another API endpoint.
Fetching the crushmap should also be quite a bit less data that needs to
be transferred, compared to the other possible nodes/<node>/ceph/osd
endpoint, especially in larger clusters.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
Ext.util.Sorter does not have an 'order' property, so 'order: DESC'
didn't have an effect. The default is 'ASC' and it is arguably the
preferred direction for all affected sorters anyways.
Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
Overriding the controller render_status must be done in an derived
class, not as config for the instantiated object.
While it worked it really isn't idiomatic and ExtJS complained with
an error in debug mode, which broke the whole UI then.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
If there is a left-over entry for a dead node in the ceph osd tree
the panel wouldn't show and produce an
Uncaught TypeError: data.versions is undefined
because of an access
node.version = data.versions[node.name];
further below (not visible in the patch itself).
AFAICT, the same issue would also happen when something went wrong
with getting the broadcasted ceph-versions, or when a node is part
of Ceph, but not PVE.
Handle the situation gracefully by always initializing data.versions.
Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
we now have to handle the version of 'old' nodes, as well as the
new nodes, which are missing the 'version' field in the metadata/osd call
so set the 'versions' object to a default empty object for osds
and get the host version directly from the treenode
(the parent node of the osd node)
also compare the host version in the dashboard by getting first the
new version field, and the old one as fallback.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
For DB and WAL disks, not only partitions will show up now, but one
more type of disk, that didn't show up before: Namely, GPT-partitioned
disks with any partitions detected as used.
It's confusing as the size shown is of the full disk, with no
indication that a new partition will be appended at the end. This
problem was already present before, but only affected GPT-partitioned
disks where no usage on a partition was detected.
Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
we want to reuse that controller type by overriding some functionality
in the future, so factor it out.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
when a daemon is stopped, the version here is 'undefined'. catch that
instead of letting the template renderer run into an error.
this fixes the rendering of the grid backgrounds
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
It seems we did not prepare the GUI enough for the API changes
planned when stopping to broadcast the old single string version. We
have to use the node specific versions, not the global 'versions'
object.
Also use the `PVE.Utils.compare_ceph_versions` everywhere, since some
versions are strings and others are the parts of the version (e.g.
["16", "2, "6"])
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
During the Ceph installation, in the configuration step, a note is shown
if there already is a Ceph configuration present. This notification
should be in the center of the wizard but is currently barely visible.
ExtJS is having trouble calculating the position and the result is that
the note is placed almost out of the visible area.
Setting a fixed height helps ExtJS calculate the vertical position.
Changing from the 'beforeshow' listener to the 'afterrender' one helps
for the horizontal position.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
we incorrectly used 'total' as 100% of the to recovered objects here,
but that containst the total number of *bytes*.
rename 'toRecover' to better reflect that the unit is 'objects' and
use that as total
reported by a user:
https://forum.proxmox.com/threads/bug-ceph-recovery-bar-not-showing-percentage.91782/
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
we should not warn for ceph's built-in default value for min_size as
having the min_size half of the size (rounded up) is ok and even the
default for ceph
Since there seems to be no 'quorum based' pg inconsistency recovery[0],
only a copy from the authoritative osd, there is nothing wrong
with setting that.
0: https://docs.ceph.com/en/latest/rados/operations/pg-repair/
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
there was a spurious leftover single quote("'"), that prevented
the correct css class to be inserted
(e.g., fa-exclamation' instead of fa-exclamation)
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Looks already OK at that size, and one gets a better overview.
We have a slightly complex layout here (to columns which should be
above each other) so we cannot just use the generic helper, but
that's OK here - it *is* a special view.
Note, not all people use full-sized windows all the time, so the
widths here must not only be considered in terms of display
resolutions...
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
