nfs locking died mysteriously, then the system went nuts

[mathog]

Yesterday NFS locking was working, today it is not. Centos 7.3.1611. The system had been up 154 days and no changes to the configuration have been made in months. NFS itself is still working.

Also noticed that after a root login over ssh "uxterm" would start with this at the top:

Code: Select all

abrt-cli status' timed out

and

Code: Select all

iptables --list

took forever. But

Code: Select all

iptables -n --list

is quick. nslookup on a variety of hosts worked.

The logs don't show anything failing, but this is suspicious:

Code: Select all

May 23 03:04:49 name yum[31266]: Updated: libtirpc.x86_64 0.2.4-0.8.el7_3
May 23 03:04:50 name yum[31266]: Updated: rpcbind.x86_64 0.2.0-38.el7_3

The NFS lock server is supposed to be locked to port 4001, but nmap shows nothing running there and 4003 open for some reason:

Code: Select all

4000/tcp  open  remoteanything
4002/tcp  open  mlchat-proxy
4003/tcp  open  pxc-splr-ft

whereas on another (nonCentos) machine configured otherwise to be very similar

Code: Select all

4000/tcp  open  remoteanything
4001/tcp  open  newoak
4002/tcp  open  mlchat-proxy

Contents of /etc/sysconfig/nfs are:

Code: Select all

LOCKD_TCPPORT=4001
LOCKD_UDPPORT=4001
RPCNFSDARGS=""
RPCNFSDCOUNT=30
RPCMOUNTDOPTS=""
MOUNTD_PORT=4002
STATDARG=" -o 4004"
STATD_PORT=4000
SMNOTIFYARGS=""
RPCIDMAPDARGS=""
RPCGSSDARGS=""
GSS_USE_PROXY="no"
RPCSVCGSSDARGS=""
BLKMAPDARGS=""

Tried to restart the nfslock demon with :

Code: Select all

systemctl restart nfs-lock.service

and that locked up. Then "ps -ef" and "top" stopped working, and I rebooted it. Probably it was running some abrt process, but I wanted to start clean at that point anyway.

When it came back up there were once again no NFS locks, but the "abrt-cli" messages were gone.

Suggestions???

[mathog]

Just noticed this thread:

viewtopic.php?f=47&t=62658

did:

Code: Select all

yum downgrade rpcbind.x86_64 0.2.0-38.el7_3 libtirpc.x86_64 0.2.4-0.8.el7_3

and nfs locking wasn't working still, but after

Code: Select all

systemctl restart nfs-lock.service

it was. However, the 4001 port still isn't shown by nmap, and the 4003 port is. The firewall has the range 4000-4004
open. Can somebody please check their Centos 7 machine and see if the expected NFS lockd port is seen by nmap?

Seems like there are some serious problems with one of those two packages!