[ClusterLabs] Solved: pacemakerd quits after few seconds with some errors

Tue Aug 23 12:56:52 EDT 2016

Found the 2 reasons:
1) I had to use gcc 4.8 for libqb to use internal memory barries
this still did not solve the crash but changed the way it crashed the subdaemons
2) /usr/var/run is not writable to everyone, but pacemakerd subdaemons want to create socket files here with hacluster user, and fail!
I will see if I can create these files in advance with correct permission during installation, but : how can I change this directory? looks like its libqb, but how can I drive this folder from the daemons? This way I could create a /usr/var/run/cluster with permissions and let everything run there.
Thanks
Gabrele
----------------------------------------------------------------------------------------
Sonicle S.r.l.
:
http://www.sonicle.com
Music:
http://www.gabrielebulfon.com
Quantum Mechanics :
http://www.cdbaby.com/cd/gabrielebulfon
Da:
Gabriele Bulfon
A:
Cluster Labs - All topics related to open-source clustering welcomed
kwenning at redhat.com
Data:
23 agosto 2016 15.05.32 CEST
Oggetto:
Re: [ClusterLabs] pacemakerd quits after few seconds with some errors
I found that pacemakerd leaves a core file where I launch it, nad here is the output from "mdb core":
sonicle at xstorage1:/sonicle/etc/cluster/corosync# mdb core
Loading modules: [ libc.so.1 ld.so.1 ]
$C
08047a48 libqb.so.0.18.0`qb_thread_lock+0x16(0, feef9875, 8047a9c, fe9eb842, fe9ff000, 806fc78)
08047a68 libqb.so.0.18.0`qb_atomic_int_add+0x22(806fd84, 1, 8047a9c, 773)
08047a88 libqb.so.0.18.0`qb_ipcs_ref+0x23(806fc78, fea30960, feef9865, fe9de139, fede608f, 806fb58)
08047ab8 libqb.so.0.18.0`qb_ipcs_create+0x68(8057fd9, 0, 0, 8069470, 805302e, 20)
08047ae8 libcrmcommon.so.3.5.0`mainloop_add_ipc_server+0x77(8057fd9, 0, 8069470, 8047b64, 0, feffb0a8)
08047b28 main+0x18e(8047b1c, fef726a8, 8047b58, 8052d2f, 1, 8047b64)
08047b58 _start+0x83(1, 8047c70, 0, 8047c8c, 8047ca0, 8047cb4)
----------------------------------------------------------------------------------------
Sonicle S.r.l.
:
http://www.sonicle.com
Music:
http://www.gabrielebulfon.com
Quantum Mechanics :
http://www.cdbaby.com/cd/gabrielebulfon
Da:
Gabriele Bulfon
A:
kwenning at redhat.com Cluster Labs - All topics related to open-source clustering welcomed
Data:
23 agosto 2016 14.30.20 CEST
Oggetto:
Re: [ClusterLabs] pacemakerd quits after few seconds with some errors
About the hacluster/haclient user/group, I staft to think that cib can't connect because it's started by pacemakerd with user hacluster, even though pacemakerd is started as root.
Instead, just before pacemakerd is able to connect with the same call, but that is the root user.
So I tried to run pacemakerd as hacluster, and infact it can't start that way.
I tried then to add the uidgid spec in the corosync.conf, but seems not to work anyway.
So ...should I start also corosync as hacluster? Is it safe to run everything as root? How can I force pacemakerd to run every child as root?
...if this is the problem...
----------------------------------------------------------------------------------------
Sonicle S.r.l.
:
http://www.sonicle.com
Music:
http://www.gabrielebulfon.com
Quantum Mechanics :
http://www.cdbaby.com/cd/gabrielebulfon
----------------------------------------------------------------------------------
Da: Klaus Wenninger
A: users at clusterlabs.org
Data: 23 agosto 2016 9.07.03 CEST
Oggetto: Re: [ClusterLabs] pacemakerd quits after few seconds with some	errors
On 08/23/2016 08:50 AM, Gabriele Bulfon wrote:
Ok, looks like Corosync now runs fine with its version, but then
pacemakerd fails again with new errors on attrd and other daemons it
tries to fork.
The main reason seems around ha signon and cluster process group api.
Any idea?
Just to be sure: You recompiled pacemaker against your new corosync?
Klaus
Gabriele
----------------------------------------------------------------------------------------
*Sonicle S.r.l. *: http://www.sonicle.com
*Music: *http://www.gabrielebulfon.com
*Quantum Mechanics : *http://www.cdbaby.com/cd/gabrielebulfon
----------------------------------------------------------------------------------
Da: Jan Pokorný
A: users at clusterlabs.org
Data: 23 agosto 2016 7.59.37 CEST
Oggetto: Re: [ClusterLabs] pacemakerd quits after few seconds with
some errors
On 23/08/16 07:23 +0200, Gabriele Bulfon wrote:
Thanks! I am using Corosync 2.3.6 and Pacemaker 1.1.4 using the
"--with-corosync".
How is Corosync looking for his own version?
The situation may be as easy as building corosync from GitHub-provided
automatic tarball, which is never a good idea if upstream has its own
way of proper release delivery:
http://build.clusterlabs.org/corosync/releases/
(specific URLs are also being part of the corosync announcements
on this list)
The issue with automatic tarballs already reported:
https://github.com/corosync/corosync/issues/116
--
Jan (Poki)
_______________________________________________
Users mailing list: Users at clusterlabs.org
http://clusterlabs.org/mailman/listinfo/users
Project Home: http://www.clusterlabs.org
Getting started:
http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
_______________________________________________
Users mailing list: Users at clusterlabs.org
http://clusterlabs.org/mailman/listinfo/users
Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
_______________________________________________
Users mailing list: Users at clusterlabs.org
http://clusterlabs.org/mailman/listinfo/users
Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
_______________________________________________Users mailing list: Users at clusterlabs.orghttp://clusterlabs.org/mailman/listinfo/usersProject Home: http://www.clusterlabs.orgGetting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdfBugs: http://bugs.clusterlabs.org
_______________________________________________Users mailing list: Users at clusterlabs.orghttp://clusterlabs.org/mailman/listinfo/usersProject Home: http://www.clusterlabs.orgGetting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdfBugs: http://bugs.clusterlabs.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20160823/15e344e4/attachment-0003.html>