Bug 1393798

Summary:	[TestOnly][platform issue] RHEL 7.3 and lockd ports setting
Product:	Red Hat Enterprise Virtualization Manager	Reporter:	CAPELLE Benoit <capelle>
Component:	ovirt-engine	Assignee:	Nikolai Sednev <nsednev>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Lukas Svaty <lsvaty>
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	4.0.4	CC:	bugs, capelle, didi, lsurette, lsvaty, mkalinin, nsednev, rbalakri, Rhev-m-bugs, sbonazzo, srevivo, ykaul, ylavi
Target Milestone:	ovirt-4.2.0	Keywords:	TestOnly
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2018-01-29 15:52:14 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	External	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Bug Depends On:	1394720
Bug Blocks:

Description CAPELLE Benoit 2016-11-10 11:08:11 UTC

Description of problem:
Red Hat released RHEL 7.3

Package nfs-utils was upgraded from nfs-utils-1.3.0-0.21.el7_2.1 to nfs-utils-1.3.0-0.33.el7

NFS lockd parameters have now to be set in a new file:
$ rpm -qi --changelog nfs-utils
...
* Mon May 16 2016 Steve Dickson <steved> 1.3.0-0.26
...
- lockd: added lockd.conf to set module parameters (bz 1264387)
...

/usr/share/ovirt-engine/setup/plugins/ovirt-engine-setup/ovirt-engine/system/nfs.py is setting NFS lockd ports in file /etc/sysconfig/nfs
LOCKD_TCPPORT=32803
LOCKD_UDPPORT=32769

/usr/share/ovirt-engine/firewalld/ovirt-engine/ovirt-nfs.xml.in is setting
NFS lockd ports
...
    <port protocol="udp" port="32769"/>
    <port protocol="tcp" port="32803"/>
...

NFS lockd won't use parameters set in /etc/sysconfig/nfs. After rebooting,
lockd is not listening on TCP port 32803 and UDP port 32769. Clients can get
stuck if trying to acquire lock as firewall is filtering ports.

Version-Release number of selected component (if applicable):
RHEL 7.3
ovirt-engine-setup-plugin-ovirt-engine-4.0.4.4-1.el7
nfs-utils-1.3.0-0.33.el7

How reproducible:
Upgrade to RHEL 7.3. Reboot and check open TCP/UDP ports

Steps to Reproduce:
1. yum update
2. reboot
3. netstat -tulnp|grep '32769\|32803'

Actual results:
null


Expected results:
tcp        0      0 0.0.0.0:32803           0.0.0.0:*               LISTEN      -                   
udp        0      0 0.0.0.0:32769           0.0.0.0:*                           - 

Additional info:
Setting NFS lockd ports in /etc/modprobe.d/lockd.conf resolves the issue:
options lockd nlm_tcpport=32803
options lockd nlm_udpport=32769

Comment 1 Yedidyah Bar David 2016-11-10 13:52:28 UTC

(In reply to CAPELLE Benoit from comment #0)
> Steps to Reproduce:
> 1. yum update
> 2. reboot
> 3. netstat -tulnp|grep '32769\|32803'

If, indeed, as it also seems to me, these are the steps to reproduce, I'd say it's a bug in nfs-utils, not oVirt.

Perhaps a very similar bug exists in oVirt, but I'd really prefer that nfs-utils fixes both - simply keep the 7.2 behavior and read conf (also) from sysconfig.

Comment 2 Yedidyah Bar David 2016-11-10 14:01:06 UTC

To clarify:

The flow for current bug is:
1. Install and setup oVirt engine on el72. Ask engine-setup to configure an NFS ISO domain.
2. yum update to el7.3
3. reboot

The flow for the "very similar bug" is:
1. Install el7.3
2. Install and setup oVirt engine on el72. Ask engine-setup to configure an NFS ISO domain.

I can think of two fixes to the former:

Fix1:
make nfs-utils reads both from sysconfig and from the new conf

Fix2:
during 'yum update nfs-utils', copy the relevant conf from sysconfig to the new conf.

If Fix1 is not used, the latter bug requires:

Fix3 (for engine-setup):
Check which nfs-utils is available, and write the conf either to sysconfig or to the new conf, depending on version.

So I suggest to do Fix1. Please close current bug and open another one on nfs-utils. Thanks.

Comment 3 Yedidyah Bar David 2016-11-10 14:43:47 UTC

(In reply to Yedidyah Bar David from comment #2)
> To clarify:
> 
> The flow for current bug is:
> 1. Install and setup oVirt engine on el72. Ask engine-setup to configure an
> NFS ISO domain.
> 2. yum update to el7.3
> 3. reboot
> 
> The flow for the "very similar bug" is:
> 1. Install el7.3
> 2. Install and setup oVirt engine on el72. Ask engine-setup to configure an
> NFS ISO domain.

Sorry, meant:
2. Install and setup oVirt engine (on el7.3). Ask engine-setup to configure an
> NFS ISO domain.

> 
> I can think of two fixes to the former:
> 
> Fix1:
> make nfs-utils reads both from sysconfig and from the new conf
> 
> Fix2:
> during 'yum update nfs-utils', copy the relevant conf from sysconfig to the
> new conf.
> 
> If Fix1 is not used, the latter bug requires:
> 
> Fix3 (for engine-setup):
> Check which nfs-utils is available, and write the conf either to sysconfig
> or to the new conf, depending on version.
> 
> So I suggest to do Fix1. Please close current bug and open another one on
> nfs-utils. Thanks.

Comment 4 Nikolai Sednev 2016-11-14 10:13:06 UTC

Cloned this bug to https://bugzilla.redhat.com/show_bug.cgi?id=1394720 as this appears a RHEL7.3 specific issue.

Comment 9 Sandro Bonazzola 2017-01-25 07:56:17 UTC

4.0.6 has been the last oVirt 4.0 release, please re-target this bug.

Comment 17 Franta Kust 2019-05-16 12:54:45 UTC

BZ<2>Jira re-sync