Project

General

Profile

Actions

Bug #39711

closed

"unable to find any IPv4 address in networks <ipv6-network>" after upgrade to nautilus on osd and mds

Added by Thore K almost 5 years ago. Updated over 3 years ago.

Status:
Duplicate
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

After upgrading our OSDs to Nautilus they refused to start logging

May 11 10:06:19 ceph-1 ceph-osd[8777]: 2019-05-11 10:06:19.131 7f5dca047d80 -1 unable to find any IPv4 address in networks '2001:abc:2100:400::/54' interfaces ''
May 11 10:06:19 ceph-1 ceph-osd[8777]: 2019-05-11 10:06:19.131 7f5dca047d80 -1 Failed to pick public address.
May 11 10:06:19 ceph-1 systemd[1]: ceph-osd@10.service: main process exited, code=exited, status=1/FAILURE
May 11 10:06:19 ceph-1 systemd[1]: Unit ceph-osd@10.service entered failed state.
May 11 10:06:19 ceph-1 systemd[1]: ceph-osd@10.service failed.
May 11 10:06:19 ceph-1 systemd[1]: ceph-osd@10.service holdoff time over, scheduling restart.
May 11 10:06:19 ceph-1 systemd[1]: Stopped Ceph object storage daemon osd.10.

Since the cluster is IPv6 only it doesn't really make sense to obtain an IPv4 address. Before the upgrade the OSDs were starting just fine using the default IPv6 address. Now we had to bind them manually in the ceph.conf which is pretty ugly.

Later during the upgrade we faced the same issue with MDS and could also "fix" it by pinning the IP-Address in the ceph.conf.


Related issues 2 (1 open1 closed)

Related to RADOS - Bug #52867: pick_address.cc prints: unable to find any IPv4 address in networks 'fd00:fd00:fd00:3000::/64' interfacesNew

Actions
Is duplicate of RADOS - Bug #46845: Newly orchestrated OSD fails with 'unable to find any IPv4 address in networks '2001:db8:11d::/120' with ms_bind_ipv6=trueResolvedMatthew Oliver

Actions
Actions #1

Updated by Nathan Cutler almost 5 years ago

  • Project changed from Ceph to Messengers
Actions #2

Updated by Dominik Csapak almost 5 years ago

i stumbled across the same issue and found that (seemingly only for osds) ms_bind_to_ipv6 does not disable ms_bind_to_ipv4
so another method would be to set ms_bind_to_ipv4 to false in the global section

Actions #3

Updated by Dominik Csapak almost 5 years ago

it is of course ms_bind_ipv6 and ms_bind_ipv4 (without the to )

Actions #4

Updated by Thore K almost 5 years ago

Thanks for sharing. This solves my issue and seems to be related to http://docs.ceph.com/docs/nautilus/rados/configuration/msgr2/#bind-configuration-options

Now what kind of issue is this?

Is it intended behaviour to prefer legacy IP and users of IP should just be warned about this (in the upgrade guide), or shouldn't this issue occur at all?

Actions #5

Updated by Matthew Oliver over 3 years ago

  • Pull request ID set to 36536

I put this PR together which addresses this issue.

Actions #6

Updated by Kefu Chai over 3 years ago

  • Is duplicate of Bug #46845: Newly orchestrated OSD fails with 'unable to find any IPv4 address in networks '2001:db8:11d::/120' with ms_bind_ipv6=true added
Actions #7

Updated by Kefu Chai over 3 years ago

  • Status changed from New to Duplicate
Actions #8

Updated by Daniel Pivonka over 2 years ago

  • Related to Bug #52867: pick_address.cc prints: unable to find any IPv4 address in networks 'fd00:fd00:fd00:3000::/64' interfaces added
Actions

Also available in: Atom PDF