[Linux-HA] hertbeat processes owners
frank at si.ct.upc.edu
Mon Nov 26 00:26:34 MST 2007
> Date: Fri, 23 Nov 2007 13:38:35 +0100
> From: Dejan Muhamedagic <dejanmm at fastmail.fm>
> Subject: Re: [Linux-HA] hertbeat processes owners
> To: General Linux-HA mailing list <linux-ha at lists.linux-ha.org>
> Message-ID: <20071123123834.GA17785 at bork.homenet>
> Content-Type: text/plain; charset=us-ascii
> On Fri, Nov 23, 2007 at 11:50:55AM +0100, Frank wrote:
>> > Hi,
>> > we are working with heartbeat 2.1.2 with the centOS rpms (on a Red Hat
>> > Linux) with 2 nodes
>> > we tried to use a custom OCF resoure from gui without success, and after
>> > consulting maillists we were informed that it was due to a bug, so we
>> > remove it from our files. Any way we are getting the same Error messages in
>> > the log files when we try to add a resource from the gui on this node :
>> > mgmtd: 2007/11/23_09:49:39 ERROR: lrm_get_rsc_class_supported(347):
>> > failed to send a rclasses message to lrmd via ch_cmd channel.
>> > we have restarted heartbeat in both nodes, reinstall it, but it's the same.
>> > May be we forgot to clean something?
> That message is really strange, because it basically says that
> the connection to the lrmd doesn't work. I'll have to investigate
> this further. Is that happening only with the resources from
> providers other than heartbeat?
After some more testing I think the problem was caused to existing files
from an old heartbeat installation (2.0.8) in
/usr/local/lib, /usr/local/bin,etc. After removing them, such message
log doesn't appear anymore. I keep on testing but
I think it is solved.
>> > One thing we are not sure that is working fine is uids on heartbeat
>> > processes:
>> > # ps -fe | grep heart
>> > root 18894 26668 0 11:26 pts/4 00:00:00 grep heart
>> > nobody 21004 21043 0 09:27 ? 00:00:35 /usr/lib/heartbeat/lrmd -r
>> > 90 21006 21043 0 09:27 ? 00:00:00 /usr/lib/heartbeat/crmd
>> > root 21043 1 0 Nov15 ? 00:00:01 heartbeat: master control
>> > process
>> > nobody 21046 21043 0 Nov15 ? 00:00:00 heartbeat: FIFO reader
>> > nobody 21047 21043 0 Nov15 ? 00:00:00 heartbeat: write: bcast eth1
>> > nobody 21048 21043 0 Nov15 ? 00:00:00 heartbeat: read: bcast eth1
>> > 90 21077 21006 0 09:28 ? 00:00:00 /usr/lib/heartbeat/tengine
>> > 90 21078 21006 0 09:28 ? 00:00:00 /usr/lib/heartbeat/pengine
>> > 90 21115 21043 0 Nov15 ? 00:00:00 /usr/lib/heartbeat/ccm
>> > 90 21116 21043 0 Nov15 ? 00:00:06 /usr/lib/heartbeat/cib
>> > nobody 21118 21043 0 Nov15 ? 00:00:00 /usr/lib/heartbeat/stonithd
>> > 90 21119 21043 0 Nov15 ? 00:00:00 /usr/lib/heartbeat/attrd
>> > root 21121 21043 0 Nov15 ? 00:00:01 /usr/lib/heartbeat/mgmtd -v
>> > process "lrmd" is owned by "nodody", but it changes to "root" sometime (???)
>> > in other node is owned by root almost all the time (90 is hacluster). Can
>> > that
>> > be causing problems?
> No. That's fine. Some processes are dropping/gaining privileges
> depending on what they are to do. In particular, lrmd and
> stonithd always run external scripts (resource/stonith agents)
> with root privileges.
Ok, excellent support. Thanks a lot Deja
Aquest missatge ha estat analitzat per MailScanner
a la cerca de virus i d'altres continguts perillosos,
i es considera que està net.
For all your IT requirements visit: http://www.transtec.co.uk
More information about the Linux-HA