[Bug 1850754] Re: ceph-volume lvm list is O(n^2)

[James Page]

This bug was fixed in the package ceph - 12.2.12-0ubuntu0.18.04.4~cloud0
---------------

 ceph (12.2.12-0ubuntu0.18.04.4~cloud0) xenial-queens; urgency=medium
 .
   * New update for the Ubuntu Cloud Archive.
 .
 ceph (12.2.12-0ubuntu0.18.04.4) bionic; urgency=medium
 .
   [ Billy Olsen ]
   * Do not validate fs caps on authorize (LP: #1847822):
     - d/p/dont-validate-fs-caps-on-authorize.patch: Do not validate
       the filesystem caps with a new client connection to the monitor
       when authorizing a client connection.
 .
   [ Dan Hill ]
   * d/p/issue38454.patch: Cherry pick of fixes for misc RGW bugs
     and cleanup of garbage collection code (LP: #1843085).
 .
   [ Dariusz Gadomski ]
   * d/p/issue37490.patch: Cherry pick fix to optimize LVM queries
     in ceph-volume, resolving performance issues in systems under
     heavy load or with large numbers of disks (LP: #1850754).


** Changed in: cloud-archive/queens
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of Ubuntu
OpenStack, which is subscribed to Ubuntu Cloud Archive.
https://bugs.launchpad.net/bugs/1850754

Title:
  ceph-volume lvm list is O(n^2)

Status in Ceph:
  Unknown
Status in Ubuntu Cloud Archive:
  Fix Released
Status in Ubuntu Cloud Archive queens series:
  Fix Released
Status in Ubuntu Cloud Archive rocky series:
  Fix Released
Status in Ubuntu Cloud Archive stein series:
  Fix Released
Status in Ubuntu Cloud Archive train series:
  Fix Released
Status in ceph package in Ubuntu:
  Fix Released
Status in ceph source package in Bionic:
  Fix Released
Status in ceph source package in Disco:
  Fix Committed
Status in ceph source package in Eoan:
  Fix Released

Bug description:
  [Impact]
  ceph-volume lvm calls take too much time and the time grows exponentially.
  A simple call may take several minutes depending on the LV configuration.

  [Test Case]
  1. Create a setup with a large number of disks and logical volumes (e.g. 24)
  2. Call ceph-volume lvm list for one of the disks.
  3. Wait for the call to finish.

  [Regression Potential]
  Patch introduces LV caching, so changing LVs while ceph-volume is running may lead to race conditions.

  [Other Info]
  Original bug description:
   
   * Anything else you think is useful to include
   * Anticipate questions from users, SRU, +1 maintenance, security teams and the Technical Board
   * and address these questions in advance

  ceph-volume lvm list calls take very long time due to the fact it has
  been implemented with O(n^2). This leads to unproportionally long
  execution time taking under consideration the fact that no complex
  computation is needed.

  Steps to reproduce:
  1. Create a setup with a large number of disks and logical volumes (e.g. 24)
  2. Call ceph-volume lvm list for one of the disks.
  3. Wait for the call to finish.

  Expected result:
  Call is executed within seconds at most.

  Actual result:
  Call may take even minutes in some circumstances (large number of disks, some OSDs stopped).

To manage notifications about this bug go to:
https://bugs.launchpad.net/ceph/+bug/1850754/+subscriptions