[pve-devel] [PATCH manager] ui: osd: warn if removal could be problematic

[pve-devel] [PATCH manager] ui: osd: warn if removal could be problematic

[Aaron Lauterer]

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.

Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
---

Those are the things to check for that came to mind. If someone thinks
that we should definitely check for more, I am happy to send a v2.

I am also open to suggestions on how to phrase the warnings better.

I opted for separate hints to be able to show detailed hints so the
users have an idea how to act on the warning and to keep the logic
behind them simple.

 www/manager6/ceph/OSD.js | 41 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)

diff --git a/www/manager6/ceph/OSD.js b/www/manager6/ceph/OSD.js
index e126f8d0..7ab11b9e 100644
--- a/www/manager6/ceph/OSD.js
+++ b/www/manager6/ceph/OSD.js
@@ -178,6 +178,20 @@ Ext.define('PVE.CephRemoveOsd', {
 	    labelWidth: 130,
 	    fieldLabel: gettext('Cleanup Disks'),
 	},
+	{
+	    xtype: 'displayfield',
+	    name: 'osd-flag-hint',
+	    userCls: 'pmx-hint',
+	    value: gettext('Global flags that limit the self healing of Ceph are enabled.'),
+	    hidden: true,
+	},
+	{
+	    xtype: 'displayfield',
+	    name: 'degraded-objects-hint',
+	    userCls: 'pmx-hint',
+	    value: gettext('Objects are degraded. Consider waiting until the cluster is healthy.'),
+	    hidden: true,
+	},
     ],
     initComponent: function() {
         let me = this;
@@ -193,6 +207,33 @@ Ext.define('PVE.CephRemoveOsd', {
 
 	me.title = gettext('Destroy') + ': Ceph OSD osd.' + me.osdid.toString();
 
+	Proxmox.Utils.API2Request({
+	    url: `/cluster/ceph/flags`,
+	    method: 'GET',
+	    failure: response => Ext.Msg.alert(gettext('Error'), response.htmlStatus),
+	    success: function({ result: { data } }) {
+		let flags = Array.from(
+		    data.filter(v => v.value),
+		    v => v.name,
+		).filter(v => ['norebalance', 'norecover', 'noout'].includes(v));
+
+		if (flags.length) {
+		    me.down('field[name=osd-flag-hint]').setHidden(false);
+		}
+	    },
+	});
+
+	Proxmox.Utils.API2Request({
+	    url: `/cluster/ceph/status`,
+	    method: 'GET',
+	    failure: response => Ext.Msg.alert(gettext('Error'), response.htmlStatus),
+	    success: function({ result: { data } }) {
+		if (Object.keys(data.pgmap).includes('degraded_objects')) {
+		    me.down('field[name=degraded-objects-hint]').setHidden(false);
+		}
+	    },
+	});
+
         Ext.applyIf(me, {
 	    url: "/nodes/" + me.nodename + "/ceph/osd/" + me.osdid.toString(),
         });
-- 
2.30.2

Re: [pve-devel] [PATCH manager] ui: osd: warn if removal could be problematic

[Thomas Lamprecht]

On 04.02.22 10:50, Aaron Lauterer wrote:
> 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.
> 
> Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
> ---
> 
> Those are the things to check for that came to mind. If someone thinks
> that we should definitely check for more, I am happy to send a v2.
> 
> I am also open to suggestions on how to phrase the warnings better.
> 
> I opted for separate hints to be able to show detailed hints so the
> users have an idea how to act on the warning and to keep the logic
> behind them simple.

independent of what we then check, this should actually but a load mask on the dialogue
or the like, as else it may be to late if backend is slow and/or network latency between
UI and PVE is high (E.g., the coverage @ deutsches eck gives me about 10 to 15s spikes
when travelling through via train, not that I expect to delete OSDs from there in
production critical setups anytime soon ;)