View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0008461||CentOS-7||pacemaker||public||2015-04-13 16:33||2017-11-29 23:05|
|Target Version||Fixed in Version|
|Summary||0008461: Pacemaker 1.1.12 <ocf:pacemaker:ping> doesn't work because of fping|
|Description||After the upgrade from Pacemaker 1.1.12 the <ocf:pacemaker:ping> resource stay at 0 weight (Connectivity lost) because it cannot write a temporary file in /tmp.|
After investigation it appears that the ping resource now use fping by default instead of ping and selinux blocks writing in /tmp.
|Steps To Reproduce||- Upgrading from CentOS 7.0 to CentOS 7.1|
- Start pacemaker cluster with pcs
- Add a cloned ping resource with pcs
- With crm_mon -Arf we can see pingd score to 0 with lost connectivity
- After putting selinux in debug mode, in audit.log we see fping can't write in /tmp
|Additional Information||<ocf:pacemaker:ping> description tells that ping is used by default whereas fping is used.|
To force fping we must use use_fping parameter
|Tags||No tags attached.|
This problem still exists in `pacemaker-cli-1.1.15-11.el7_3.5.x86_64` under CentOS 7.4.
As a note for others who may run into this problem before RedHat decides to fix it, if you add `use_fping=0` to your "pcs resource create" command, the resource agent script will *not* use `fping`, and the resource seems to work as expected.
|Was this ever reported upstream on bugzilla.redhat.com? If not then it will never get fixed as pacemaker in CentOS is a direct rebuild from the same source that is used for RHEL.|