[Pacemaker] Pacemaker 0.7.3: How to use pingd clone resource and constraints
Andrew Beekhof
beekhof at gmail.com
Fri Sep 26 10:42:57 EDT 2008
On Fri, Sep 26, 2008 at 09:37, Andrew Beekhof <beekhof at gmail.com> wrote:
> Most likely you've found a bug :-(
> Would you be able to create a bugzilla entry for this?
> On Sep 25, 2008, at 9:33 PM, Bruno Voigt wrote:
>
> Wow.. these warnings are even shown for 127.0.0.1 ?!
> Do I need to finetune the IP stack options somewhere like in sysctl.conf
> to have these warnings of pingd fixed?
I looked into this, and for localhost the cause is me not filtering
out the ICMP_ECHO packet.
I'll patch that now.
For remote hosts, there seems to be something else going on
> root at xen20b:~# /usr/lib/heartbeat/pingd -V -a pingd-internal -d 5s -m 1000
> -h 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:04 debug: main: Adding ping host 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:04 debug: main: attrd registration attempt: 0
> pingd[26741]: 2008/09/25_21:30:09 debug: init_client_ipc_comms_nodispatch:
> Attempting to talk on: /var/run/heartbeat/crm/attrd
> pingd[26741]: 2008/09/25_21:30:09 info: main: Starting pingd
> pingd[26741]: 2008/09/25_21:30:19 debug: stand_alone_ping: Checking
> connectivity
> pingd[26741]: 2008/09/25_21:30:19 debug: ping_open: Got address 127.0.0.1
> for 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:19 debug: ping_open: Opened connection to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:19 debug: ping_write: Sent 39 bytes to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:19 WARN: dump_v4_echo: Bad echo (0): 8,
> code=0, seq=5, id=0, check=22763
> pingd[26741]: 2008/09/25_21:30:20 debug: ping_write: Sent 39 bytes to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:20 debug: dump_v4_echo: 59 bytes from
> 127.0.0.1, icmp_seq=5: beekhof-v4
> pingd[26741]: 2008/09/25_21:30:21 debug: ping_write: Sent 39 bytes to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:21 WARN: dump_v4_echo: Bad echo (0): 8,
> code=0, seq=13138, id=0, check=3000
> pingd[26741]: 2008/09/25_21:30:22 debug: ping_write: Sent 39 bytes to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:22 debug: dump_v4_echo: 59 bytes from
> 10.10.10.167, icmp_seq=13138: beekhof-v4
> pingd[26741]: 2008/09/25_21:30:23 debug: ping_write: Sent 39 bytes to
> 127.0.0.1
> pingd[26741]: 2008/09/25_21:30:23 WARN: dump_v4_echo: Bad echo (0): 8,
> code=0, seq=8284, id=0, check=459
>
> root at xen20b:~# uname -a
> Linux xen20b.bb.ic3s.de 2.6.18-6-xen-amd64 #1 SMP Tue Aug 19 06:15:09 UTC
> 2008 x86_64 GNU/Linux
>
> Bruno Voigt wrote:
>
> Hi Andrew,
>
> is pingd doing alive tests differently compared to the normal ping command?
> normal & flood ping of these hosts show 0% packet lost from my both nodes.
>
> In the log below pingd - besides the warnings -
> states that the node is alive and that it had sent an update,
> but it does not show up in the cib.
>
> WR,
> Bruno
>
> Andrew Beekhof wrote:
>
>
> On Sep 24, 2008, at 10:36 PM, Serge Dubrouski wrote:
>
>
>
> There is a problem with attrd that affects pingd in Pacemeaker
> 0.7.3/Heartbeat 2.99. I've already created a Bugzilla ticket for it.
> You can add your information there:
>
> http://developerbugs.linux-foundation.org/show_bug.cgi?id=1969
>
>
> I'm not so sure this is the same thing.
> Those "Bad echo" messages look suspicious
>
>
>
> Sep 24 22:01:46 xen20a pingd: [13142]: WARN: dump_v4_echo: Bad echo
> (0):
> 3, code=1, seq=0, id=0, check=24031
>
>
> In fact, type=3 is ICMP_DEST_UNREACH - so pingd really is having
> trouble contacting that host.
>
>
>
>
> On Wed, Sep 24, 2008 at 2:04 PM, Bruno Voigt <Bruno.Voigt at ic3s.de>
> wrote:
>
>
> I defined two ping clone resources,
> to be used independently by different resources:
>
> <clone id="clone-pingd-internal">
> <primitive id="pingd-internal" provider="pacemaker" class="ocf"
> type="pingd">
> <instance_attributes id="pingd-internal-ia">
> <nvpair id="pingd-internal-ia01" name="name"
> value="pingd-internal"/>
> <nvpair id="pingd-internal-ia02" name="dampen" value="5s"/>
> <nvpair id="pingd-internal-ia03" name="multiplier"
> value="1000"/>
> <nvpair id="pingd-internal-ia04" name="host_list"
> value="172.17.32.23 192.168.132.23"/>
> </instance_attributes>
> </primitive>
> </clone>
>
> <clone id="clone-pingd-external">
> <primitive id="pingd-external" provider="pacemaker" class="ocf"
> type="pingd">
> <instance_attributes id="pingd-external-ia">
> <nvpair id="pingd-external-ia01" name="name"
> value="pingd-external"/>
> <nvpair id="pingd-external-ia02" name="dampen" value="5s"/>
> <nvpair id="pingd-external-ia03" name="multiplier"
> value="1000"/>
> <nvpair id="pingd-external-ia04" name="host_list"
> value="195.244.97.241"/>
> </instance_attributes>
> </primitive>
> </clone>
>
> I defined a constraint for a resource so that it depends on
> pingd-internal
>
> <constraints>
> <rsc_location id="hbtest1b-connectivity" rsc="hbtest1b">
> <rule id="hbtest1b-connectivity-exclude-rule" score="-INFINITY" >
> <expression id="hbtest1b-connectivity-exclude"
> attribute="pingd-internal" operation="not_defined"/>
> </rule>
> </rsc_location>
> </constraints>
>
> But this causes the resource to be unrunnable on either of my both
> nodes,
>
>
> There are as expected 2 pingd daemons running:
>
> root 6132 1 0 21:07 ? 00:00:00
> /usr/lib/heartbeat/pingd
> -D -p /var/run/heartbeat/rsctmp/pingd-pingd-internal:0 -a
> pingd-internal
> -d 5s -m 1000 -h 172.17.32.23 -h 192.168.132.23
> root 13142 1 0 21:47 ? 00:00:00
> /usr/lib/heartbeat/pingd
> -D -p /var/run/heartbeat/rsctmp/pingd-pingd-external:0 -a
> pingd-external
> -d 5s -m 1000 -h 195.244.97.241
>
> The problem is, I can't see in the cibadmin -Q output that the pingd
> daemons have
> have stored their results anywhere..
>
> In the log I see the following output:
>
> Sep 24 22:01:46 xen20a pingd: [13142]: WARN: dump_v4_echo: Bad echo
> (0):
> 3, code=1, seq=0, id=0, check=24031
> Sep 24 22:01:47 xen20a pingd: [13142]: WARN: dump_v4_echo: Bad echo
> (0):
> 3, code=1, seq=0, id=0, check=24031
> Sep 24 22:01:48 xen20a pingd: [13142]: WARN: dump_v4_echo: Bad echo
> (0):
> 8, code=0, seq=261, id=0, check=22762
> Sep 24 22:01:48 xen20a pingd: [6132]: WARN: dump_v4_echo: Bad echo (0):
> 8, code=0, seq=263, id=0, check=22250
> Sep 24 22:01:50 xen20a pingd: [13142]: info: stand_alone_ping: Node
> 195.244.97.241 is alive (1)
> Sep 24 22:01:50 xen20a pingd: [13142]: info: send_update: 1 active ping
> nodes
> Sep 24 22:01:51 xen20a pingd: [6132]: WARN: dump_v4_echo: Bad echo (0):
> 3, code=1, seq=0, id=0, check=24031
> Sep 24 22:01:52 xen20a pingd: [6132]: info: stand_alone_ping: Node
> 172.17.32.23 is alive (3)
> Sep 24 22:01:53 xen20a pingd: [6132]: WARN: dump_v4_echo: Bad echo (0):
> 3, code=1, seq=0, id=0, check=24031
> Sep 24 22:01:54 xen20a pingd: [6132]: WARN: dump_v4_echo: Bad echo (0):
> 3, code=1, seq=0, id=0, check=25823
> Sep 24 22:01:55 xen20a pingd: [6132]: WARN: dump_v4_echo: Bad echo (0):
> 3, code=1, seq=0, id=0, check=24031
> Sep 24 22:01:57 xen20a pingd: [6132]: info: stand_alone_ping: Node
> 192.168.132.23 is alive (2)
> Sep 24 22:01:57 xen20a pingd: [6132]: info: send_update: 2 active
> ping nodes
>
> Where should the current pingd status be located in the cib ?
> What is wrong with my setup ?
>
> TIA,
> Bruno
>
>
>
> --
> Serge Dubrouski.
>
>
>
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at clusterlabs.org
> http://list.clusterlabs.org/mailman/listinfo/pacemaker
>
>
More information about the Pacemaker
mailing list