[pve-devel] applied: [PATCH manager 00/15] improve ceph dashboard

[Thomas Lamprecht]

On 5/27/19 2:13 PM, Dominik Csapak wrote:
> this series aims to improve the ceph dashbaord and make it cluster-wide
> it is a continuation of my rfc from april[0]
> 
> i introduce 2 cluster wide api calls for status/metadata
> which are used in the dashboard to primarily show outdated services
> 
> it uses the broadcast_node_kv interface from Cluster.pm to
> broadcast and read ceph versions and ceph service unit existence from
> each host
> 
> the dashboard itself got a reorganization:
> * the services (mon/mgr/mds) now have a seperate panel
> * outdated services (osd/mon/mgr/mds) are highlighted
> * the list of pgs is now scrollable
> * there is a new pie chart for the rough pg state (like ceph mgr dashboard)
>   to quickly see the overall pg state
> * the service tooltips are vastly improved (icons/better formatting/etc.)
> 
> what is yet to come for the gui:
> 
> * better control (start/stop/restart) for all services (mon/mgr/mds)
> * version info in the specific panels (osd/mon/mgr/mds)
> * additional osd control (e.g., scrub, more global flags)
> * display nautilus db under config
> * add more panels to datacenter (all that make sense)
> 
> 0: https://pve.proxmox.com/pipermail/pve-devel/2019-April/036739.html
> 
> Dominik Csapak (15):
>   add get_local_services for ceph
>   broadcast ceph service data to cluster
>   add cluster wide ceph api calls
>   gui: StateProvider: add fireevent option to set
>   gui: ceph: add 'old' to ceph health
>   gui: ceph: add get_ceph_icon_html to utils
>   gui: ceph: add Services.js
>   gui: ceph: remove mon specific code from StatusDetail
>   gui: ceph: prepare statusdetail for metadata
>   gui: ceph: make status panel node independant
>   gui: ceph: add metadata api call to dashboard
>   gui: ceph: add pg pie chart
>   gui: ceph: add Services panel
>   gui: ceph: tweak heights
>   gui: add datacenter ceph option
> 
>  PVE/API2/Cluster.pm               | 115 ++++++++++++
>  PVE/Ceph/Services.pm              |  18 ++
>  PVE/Service/pvestatd.pm           |  28 +++
>  www/manager6/Makefile             |   1 +
>  www/manager6/StateProvider.js     |   4 +-
>  www/manager6/Utils.js             |  28 +++
>  www/manager6/ceph/Services.js     | 374 ++++++++++++++++++++++++++++++++++++++
>  www/manager6/ceph/Status.js       |  76 ++++++--
>  www/manager6/ceph/StatusDetail.js | 316 +++++++++++++++++---------------
>  www/manager6/dc/Config.js         |   6 +
>  www/manager6/dc/Health.js         |   9 +-
>  11 files changed, 803 insertions(+), 172 deletions(-)
>  create mode 100644 www/manager6/ceph/Services.js
> 

applied series, but the "03/15 add cluster wide ceph api calls" had a perl
"compilation error", i.e., hash access to a hash:

-                   for my $k (keys %$data->{$id}) {
+                   for my $k (keys %{$data->{$id}}) {


I took the freedom of directly squashing this into your patch, FYI.

also two minor opinionated cleanups patches in between. Thanks!