[Linux-HA] strange monitor behaviour

[Pavol Gono]

On 1/4/07, Andrew Beekhof <beekhof at gmail.com> wrote:
> grumble... you go on holidays and look what happens :-(
>
> i'll take a look at this tomorrow

I made the same procedure on another two machines (mach11s10,
mach13s10), distros are SLES10  (debo was debian, fico was gentoo). I
used the sources from http://hg.linux-ha.org/dev
changeset 9918 (latest commit at Tue, 02 Jan 2007 11:33:56 +0100).
Configure options: --with-group-id=90 --with-ccmuser-id=90
CFLAGS='-fno-unit-at-a-time -g -O0' --sysconfdir=/etc
--localstatedir=/var --enable-bundled_ltdl --enable-ltdl-convenience
--disable-fatal-warnings --disable-tipc --disable-ldirectord
--disable-snmp --disable-mgmt --disable-quorumd --enable-crm-dev

After start of HBs, resources appeared on mach13s10, I removed /tmp/a,
resource Dummy was stopped on mach13s10, no failover. (see attachment)
The difference between previous test and this one: HB didn't try to
start Dummy after monitor failure. After stop of heartbeat on
mach13s10, IPaddr resource was correctly stopped.

After Dummy was stopped, HB set failcount. I tried to remove failcout
with command
crm_failcount -VVV -D -U mach11s10 -r x_Dummy
It was successful (fail-count-x_Dummy property was removed), but it
wrote 2 strange errors to log:
crm_failcount[16557]: 2007/01/04_16:14:54 info: main: Mapped mach11s10
to e0ab5c91-eccf-49e2-b966-5ab3cf3ec6da
crm_failcount[16557]: 2007/01/04_16:14:54 ERROR:
cib_native_perform_op: Call failed: The object/attribute does not
exist
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: crm_log_message_adv:
#=========  message start ==========#
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG: Dumping message
with 6 fields
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[0] : [t=cib]
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[1] : [cib_clientid=16557]
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[2] : [cib_callopt=1052672]
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[3] : [cib_callid=6]
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[4] : [cib_op=cib_modify]
crm_failcount[16557]: 2007/01/04_16:14:54 WARN: MSG[5] : [cib_rc=-22]
crm_failcount[16557]: 2007/01/04_16:14:54 ERROR: update_attr: Error
setting last-lrm-refresh=1167923694 (section=crm_config,
set=cib-bootstrap-options): The object/attribute does not exist
crm_failcount[16557]: 2007/01/04_16:14:54 info: log_data_element:
update_attr: Update <cluster_property_set id="cib-bootstrap-options">
crm_failcount[16557]: 2007/01/04_16:14:54 info: log_data_element:
update_attr: Update   <attributes>
crm_failcount[16557]: 2007/01/04_16:14:54 info: log_data_element:
update_attr: Update     <nvpair
id="cib-bootstrap-options-last-lrm-refresh" name="last-lrm-refresh"
value="1167923694"/>
crm_failcount[16557]: 2007/01/04_16:14:54 info: log_data_element:
update_attr: Update   </attributes>
crm_failcount[16557]: 2007/01/04_16:14:54 info: log_data_element:
update_attr: Update </cluster_property_set>

Such error messages of crm_failure are common also with other HB configurations.

Palo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: strange_monitor_behaviour2.tar.bz2
Type: application/x-bzip2
Size: 11248 bytes
Desc: not available
Url : http://lists.community.tummy.com/pipermail/linux-ha/attachments/20070104/e7b23cac/strange_monitor_behaviour2.tar.bin