Ceph osd df including block.db size

[Dyweni - Ceph-Users <6EXbab4FYk8H@xxxxxxxxxx>]

Hi,

I realize this comes from an end-of-life release (Mimic), but I believe 
it still applies to current releases.

I've just upgraded from Luminous to Mimic and noticed that my 'ceph osd 
df' output now includes the block.db sizes.  This is pushing all OSD's 
into nearfull status (~70% used in Luminous, ~85% used in Mimic).  Each 
OSD is 1TB HDD (block) and 1TB NVME (block.db).

I'd like to understand two things:
1. Why is the reasoning behind including the block.db size/usage in 
'ceph osd df'?
2. Is there a way to exclude block.db size/usage from 'ceph osd df'?

I've looked through the ceph git comments and noticed the following:

* b97f3d6761208a1d317db3d0dba8371181e0b49f (8/21/2016, v11.0.0/master), 
Sage added:
+    // include dedicated db, too, if that isn't the shared device.
+    if (bluefs_shared_bdev != BlueFS::BDEV_DB) {
+      buf->available += bluefs->get_free(BlueFS::BDEV_DB);
+      buf->total += bluefs->get_total(BlueFS::BDEV_DB);
+    }

* f798b13a043523e1677abdcb3df0d9ced8774098 (1/31/2018, v13.0.1/master), 
Igor removed:
-    // include dedicated db, too, if that isn't the shared device.
-    if (bluefs_shared_bdev != BlueFS::BDEV_DB) {
-      buf->total += bluefs->get_total(BlueFS::BDEV_DB);
-    }

* 266d1649cf1d57ab32a71a39347bb418a4490c78 (2/20/2018, v13.0.1/master), 
Sage re-added:
+  if (bluefs) {
+    // include dedicated db, too, if that isn't the shared device.
+    if (bluefs_shared_bdev != BlueFS::BDEV_DB) {
+      buf->total += bluefs->get_total(BlueFS::BDEV_DB);
     }

I've checked the main branch (as of right now, commit 
d622325c17bb98672ad22e60a372becfbeb04cff) and the calculation still 
exists:
  File: src/os/bluestore/BlueStore.cc
  Line: 11480
    // include dedicated db, too, if that isn't the shared device.
    if (bluefs_layout.shared_bdev != BlueFS::BDEV_DB) {
      buf->total += bluefs->get_total(BlueFS::BDEV_DB);
    }

Thanks,
Dyweni
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx