[ClusterLabs] Antw: Re: Q: ocf:pacemaker:NodeUtilization monitor

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Wed May 29 04:12:55 EDT 2019


>>> Jan Pokorný <jpokorny at redhat.com> schrieb am 28.05.2019 um 16:31 in
Nachricht
<20190528143145.GA29171 at redhat.com>:
> On 27/05/19 08:28 +0200, Ulrich Windl wrote:
>> I copnfigured ocf:pacemaker:NodeUtilization more or less for fun, and I 
> realized that the cluster rrepiorts no problems, but in syslog I have these

> unusual messages:
>> 2019‑05‑27T08:21:07.748149+02:00 h06 lrmd[16599]:   notice: 
> prm_node_util_monitor_300000:15028:stderr [ info: Writing node (dir)Top...
]
>> 2019‑05‑27T08:21:07.748546+02:00 h06 lrmd[16599]:   notice: 
> prm_node_util_monitor_300000:15028:stderr [ info: Cannot find node `(dir)GNU

> Free Documentation License'. ]
>> 2019‑05‑27T08:21:07.748799+02:00 h06 lrmd[16599]:   notice: 
> prm_node_util_monitor_300000:15028:stderr [ info: Done. ]
>> 
>> 
>> "(dir)" looks a lot like Documentation. What has the monitor to do with 
> documentation?
> 
> The above looks as if you run "info" without arguments (it will try
> to dispay initial page '(dir)Top' ‑‑ and moreover perhasp when it is
> not found).
> 
> I have no idea how this could happen, since there's the only reference
> to "info" and it seems basic‑sanity guarded:
> 
>
https://github.com/ClusterLabs/resource‑agents/blob/v4.2.0/heartbeat/NodeUtil

> ization#L119
> 
>> 118     if [ ‑x $xentool ]; then
>> 119         $xentool info | awk
'/total_memory/{printf("%d\n",$3);exit(0)}'
>> 120     else
>> 121         ocf_log warn "Can only set hv_memory for Xen hypervisor"
>> 122         echo "0"
> 
> So kind of a mystery :‑)

Except when $xentool is undefined ;-)
Actually on my system this command creates an empty line:
echo $(which xl 2> /dev/null || which xm 2> /dev/null)


My configuration is:
# crm configure show prm_node_util
primitive prm_node_util ocf:pacemaker:NodeUtilization \
        op start interval=0 timeout=90 \
        op stop interval=0 timeout=120 \
        op monitor interval=300 timeout=90

> 
>> Despite of that the RAS seems to work. (SLES12 SP4 with current
>> patches applied)
> 
> Are you sure the resource's provider is pacemaker and not heartbeat?

Genuine SUSE:
# rpm -qf /usr/lib/ocf/resource.d/pacemaker/NodeUtilization
pacemaker-cli-1.1.19+20181105.ccd6b5b10-3.10.1.x86_64
# ll /usr/lib/ocf/resource.d/pacemaker/NodeUtilization
lrwxrwxrwx 1 root root 28 Apr 29 14:58
/usr/lib/ocf/resource.d/pacemaker/NodeUtilization ->
../heartbeat/NodeUtilization

> Got stuck for a bit trying to look up that agent on pacemaker side
> (booth is admittedly also a bit misleading in this regard).
> 
> ‑‑ 
> Jan (Poki)





More information about the Users mailing list