Re: What does the ceph-volume@simple-crazyhexstuff SystemD service do? And what to do about oversized MDS cache?

Paul Emmerich <paul.emmerich@xxxxxxxx> · Thu, 5 Dec 2019 13:25:26 +0100

The ceph-volume services make sure that the right partitions are mounted at /var/lib/ceph/osd/ceph-X

In "simple" mode the service gets the necessary information from a json file (long-hex-string.json) in /etc/ceph

ceph-volume simple scan/activate create the json file and systemd unit.

ceph-disk used udev instead for the activation which was *very* messy and a frequent cause of long startup delays (seen > 40 minutes on encrypted ceph-disk OSDs)

Paul

-- 
Paul Emmerich

Looking for help with your Ceph cluster? Contact us at https://croit.io

croit GmbH
Freseniusstr. 31h
81247 München
www.croit.io
Tel: +49 89 1896585 90

On Thu, Dec 5, 2019 at 1:03 PM Ranjan Ghosh <ghosh@xxxxxx> wrote:

    Hi all,
    After upgrading to Ubuntu 19.10 and consequently from Mimic to
      Nautilus, I had a mini-shock when my OSDs didn't come up. Okay, I
      should have read the docs more closely, I had to do:

    # ceph-volume simple scan /dev/sdb1

    # ceph-volume simple activate --all

    Hooray. The OSDs came back to life. And I saw that some weird
      services were created. Didn't give that much thought at first, but
      later I noticed there is now a new service in town:
    ===

    root@yak1 ~ # systemctl status ceph-volume@simple-0-6585a10b-917f-4458-a464-b4dd729ef174.service

      ceph-volume@simple-0-6585a10b-917f-4458-a464-b4dd729ef174.service
      - Ceph Volume activation:
      simple-0-6585a10b-917f-4458-a464-b4dd729ef174

         Loaded: loaded (/lib/systemd/system/ceph-volume@.service;
      enabled; vendor preset: enabled)

         Active: inactive (dead) since Wed 2019-12-04 23:29:15 CET; 13h
      ago

       Main PID: 10048 (code=exited, status=0/SUCCESS)
    ===
    Hmm. It's dead. But my cluster is alive & kicking, though.
      Everything is working. Why is this needed? Should I be worried? Or
      can I safely delete that service from /etc/systemd/... since it's
      not running anyway?
    Another, probably minor issue: 

    I still get a HEALTH_WARN "1 MDSs report oversized cache". But it
      doesn't tell me any details and I cannot find anything in the
      logs. What should I do to resolve this? Set
      mds_cache_memory_limit? How do I determine an acceptable value?

    Thank you / Best regards

    Ranjan

_______________________________________________

ceph-users mailing list

ceph-users@xxxxxxxxxxxxxx

http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com