Project

General

Profile

Feature #23003

Throw a warning/notice when --journal is specified with --bluestore

Added by David Galloway about 1 year ago. Updated about 1 year ago.

Status:
Resolved
Priority:
Normal
Assignee:
Target version:
-
Start date:
02/14/2018
Due date:
% Done:

0%

Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Pull request ID:

Description

I'm unable to tell whether an OSD I created with ceph-volume is actually using the logical volume I specified for the journal. This information used to be displayed with ceph-disk list.

https://arvimal.blog/2015/08/05/how-to-check-the-journal-disk-for-any-particular-osd/

History

#1 Updated by David Galloway about 1 year ago

  • Subject changed from Show journal information to Throw a warning/notice when --journal is specified with --bluestore

I'm changing this to a suggestion since ceph-volume /does/ display journal information. ceph noobs like me might benefit from being notified that bluestore OSDs don't have journals.

#2 Updated by Alfredo Deza about 1 year ago

  • Status changed from New to In Progress

I have a work in progress to address this, but it is non-trivial as there isn't support in argparse to express a group of flags that are incompatible with a second group of flags.

It merely has support for just a simple XOR implementation, that is: from a group of flags, only one can be chosen (https://docs.python.org/2/library/argparse.html#mutual-exclusion)

The fix will poke at the argparse object, but it will not support anything that is not passed in (e.g. default flags).

#3 Updated by Alfredo Deza about 1 year ago

  • Status changed from In Progress to Need More Info

David: given that there is no actual support for this in argparse, and I am not fully confident in my implementation, would a warning rather than an error suffice? In the past, for other
tools that give lots of output, it might not be visible enough to just send out warnings.

But a warning that might not be seen (but still be present in the output) is better than stopping on a false positive or not giving anything back at all.

Would you be OK with just a warning in this case?

#4 Updated by David Galloway about 1 year ago

Alfredo Deza wrote:

Would you be OK with just a warning in this case?

Absolutely -- especially given the fact that the message is being displayed due to my user error.

#5 Updated by Alfredo Deza about 1 year ago

  • Status changed from Need More Info to Need Review
  • Assignee set to Alfredo Deza

#6 Updated by Alfredo Deza about 1 year ago

  • Status changed from Need Review to Resolved

merged commit a09b319 into master

Also available in: Atom PDF