patch for multipath disks

[Giuseppe Attardi]

According to this page:

 https://patchwork.kernel.org/patch/9826353/

there is a bug in ceph that affects the use of FibreChannel disks.

The following patch to file hooks/ceph_hooks.py, fixes it:

*** hooks/ceph_hooks.py 2018-02-02 16:15:40.304388602 +0100
--- hooks/ceph_hooks.py~ 2018-02-02 16:18:47.304401004 +0100
***************
*** 369,378 ****

 @hooks.hook('storage.real')
 def prepare_disks_and_activate():
- # patch for missing dm devices
- # see: https://patchwork.kernel.org/patch/9826353/
- patch_persistent_storage_rules()
-
     osd_journal = get_journal_devices()
     check_overlap(osd_journal, set(get_devices()))
     log("got journal devs: {}".format(osd_journal), level=DEBUG)
--- 369,374 ----
***************
*** 558,579 ****
     log('Updating status.')

- # patch for missing dm devices
- from subprocess import check_call
-
-
- CEPH_PERSITENT_STORTAGE_RULES = '/lib/udev/rules.d/60-persistent-storage.rules'
-
-
- def patch_persistent_storage_rules():
- if os.path.isfile(CEPH_PERSITENT_STORTAGE_RULES):
- check_call(['sed', '-i', 's/KERNEL!="loop/KERNEL!="dm*|loop/',
- CEPH_PERSITENT_STORTAGE_RULES])
- log('Patched %s' % CEPH_PERSITENT_STORTAGE_RULES)
- else:
- log('Missing %s' % CEPH_PERSITENT_STORTAGE_RULES)
-
-
 if __name__ == '__main__':
     try:
         hooks.execute(sys.argv)
--- 554,559 ——

[James Page (james-page)]

Ok so the charm is a convenient place to hotfix this for an impacted deployment, but the right place to fix this is actually in the udev package so switching the bug task to the right target.