Ceph env fails

Bug #1261711 reported by Andrey Korolyov
10
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Invalid
High
Andrey Korolyov

Bug Description

Simple env, three OSD and one controller

ceph -s
  cluster bff99f2e-4daf-4653-9c30-22db84d55a7a
   health HEALTH_OK
   monmap e1: 1 mons at {node-4=192.168.0.2:6789/0}, election epoch 2, quorum 0 node-4
   osdmap e11: 2 osds: 2 up, 2 in
    pgmap v40: 492 pgs: 492 active+clean; 0 bytes data, 4169 MB used, 41260 MB / 45429 MB avail
   mdsmap e1: 0/0/1 up

third OSD node failed, check the logs

seemingly it`s our logic problem (no third osd, rerun of ceph-deploy)

Tags: ceph
Revision history for this message
Andrey Korolyov (xdeller) wrote :
  • fail Edit (3.6 MiB, application/x-tar)
Changed in fuel:
importance: Undecided → High
assignee: nobody → Dmitry Borodaenko (dborodaenko)
description: updated
Changed in fuel:
milestone: none → 4.0
Andrew Woodward (xarses)
Changed in fuel:
status: New → Triaged
tags: added: ceph
Revision history for this message
Andrew Woodward (xarses) wrote :

From node-7 /root/ceph.log

2013-12-17 11:23:14,726 [ceph_deploy.cli][INFO ] Invoked (1.2.7): /usr/bin/ceph-deploy --overwrite-conf config pull node-4
2013-12-17 11:23:14,726 [ceph_deploy.config][DEBUG ] Checking node-4 for /etc/ceph/ceph.conf
2013-12-17 11:23:14,727 [ceph_deploy.sudo_pushy][DEBUG ] will use a remote connection without sudo
2013-12-17 11:23:17,733 [ceph_deploy.config][ERROR ] Unable to pull /etc/ceph/ceph.conf from node-4
2013-12-17 11:23:17,733 [ceph_deploy][ERROR ] GenericError: Failed to fetch config from 1 hosts

Implies broken network or failed ssh keys

Revision history for this message
Dmitry Borodaenko (angdraug) wrote :

Management network was misconfigured on node-7: it wasn't able to connect to the controller (node-4) at 192.168.0.2, so it failed to pull ceph.conf, blocking the rest of ceph-osd deployment.

Revision history for this message
Dmitry Borodaenko (angdraug) wrote :

node-7 was patched differently from other nodes: admin network is patched to eth2 instead of eth0, and OVS br-mgmt and br-storage bridges were assigned to eth1 instead of eth2.

Changed in fuel:
assignee: Dmitry Borodaenko (dborodaenko) → Andrey Korolyov (xdeller)
Revision history for this message
Mike Scherbakov (mihgen) wrote :

Andrey,
your title is not descriptive enough. Please provide better title so we can understand what the bug is about reading just title.

Revision history for this message
Andrey Korolyov (xdeller) wrote :

It`s all about non-strict networking checking in Fuel. May be it`s necessary to check connectivity just after Deploy button was hit and throw a warning. Of course new modal window should be introduced.

Changed in fuel:
status: Triaged → Invalid
Revision history for this message
Andrew Woodward (xarses) wrote :

I created https://bugs.launchpad.net/fuel/+bug/1261940 to help prevent this kind of issue.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.