starlingx/integ
Code Issues Proposed changes

Ensure loading order for nfscommon and mountnfs

Browse Source 
In some rare occasions following a host lock/unlock or force
reboot, there are two rpc.statd process one of which is a
runaway that floods daemon.log causing high CPU alarm. Based
on a few samples of startup sequence (success vs failure),
it is evident that the issue appeared each time mountnfs unit
had been loaded before rpc.statd process was started.

This change ensures nfscommon unit, which starts rpc.statd
and rpc.idmapd, is loaded before mountnfs unit.

Tests performed:
  - installation (standard system)
  - sanity (automated, repeats = 10)

Closes-Bug: 1794366
Change-Id: I6a916b9c3348338b0ab79fcec4f2cd5d04741499
Signed-off-by: Tee Ngo <Tee.Ngo@windriver.com>
This commit is contained in:
Tee Ngo 2018-10-26 14:40:06 -04:00
parent b9d034298f
commit d7d8b3cf06
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
base/initscripts/centos/files/mountnfs.service
View File

@ -1,6 +1,6 @@
[Unit] [Unit]
Description=Titanium Cloud Filesystem Auto-mounter Description=Titanium Cloud Filesystem Auto-mounter
After=network.target After=network.target nfscommon.service
Before=uexportfs.service Before=uexportfs.service
[Service] [Service]

2
filesystem/nfs-utils/files/nfscommon.service
View File

@ -1,7 +1,7 @@
[Unit] [Unit]
Description=Titanium Cloud Filesystem Common Description=Titanium Cloud Filesystem Common
After=network.target rpcbind.service After=network.target rpcbind.service
Before=nfsserver.service Before=nfsserver.service mountnfs.service
[Service] [Service]
Type=oneshot Type=oneshot