Oracle Communications Performance Intelligence Center (PIC) Software
 

PLA-Support>Sun Systems>CommsGBU>Global Signaling Solutions>SN-SND: Tekelec PIC
 

In this Document

Symptoms

Changes

Cause

Solution

Procedure Overview

PDU Storage

Dataexport

Detailed file modifications

PDU Storage

Oracle Linux 6.6, 7.0 and 7.1

TPD and Oracle Linux 7.2

Dataexport

Created from <SR 3-13499781284>

Applies to:

Symptoms

IXP server hangs and shows out of memory messages.

Following can also be logged:

• BMC IPMI Watchdog Timer Timeout: Action=System Power Reset, TimerUse: 0x44, timerActions: 0x01
• in /var/log/message, oom-killer lines

Changes

Cause

There is a bug with nfsv4.

Any nfs mount point can be the cause (either PDU or dataexport).

Run following command to determine which mount points are using nfsv4:

Solution

NFS version must be changed for PDU servers and dataexport mounts or system can be upgraded to PIC 10.2 or later.

Procedure Overview

PDU Storage

1. Modify NFS configuration file on:
   1. PDU server if using Oracle Linux 7.0 or 7.1
   2. Base server writing to PDU server if using Oracle Linux 7.2 or TPD
2. During maintenance window, reboot all servers of subsystem in following order:
   1. All PDU servers
   2. All base servers writing to PDU server
3. After all servers are up, verify build log

Dataexport

1. Modify dataexport.properties
2. Restart DataExport process

Detailed file modifications

PDU Storage

Oracle Linux 6.6, 7.0 and 7.1

To prevent NFSv4 connection to PDU servers:

1. If TPD based, check out file:
   
   rcstool co /etc/sysconfig/nfs
2. Edit file /etc/sysconfig/nfs
3. Change line:
   
   RPCNFSDARGS=""
   to
   
   RPCNFSDARGS="--no-nfs-version 4"
4. If TPD based, check in file:
   
   rcstool ci /etc/sysconfig/nfs

This will have to be applied on all PDU servers and all servers must be rebooted (including all servers writing to PDU servers).

TPD and Oracle Linux 7.2

Because of a bug on OL2 to take into account RPCNFSDARGS, following workaround must be applied instead on all servers writing to PDU servers (usually all IXP base servers).

1. Check out fstab file:
   
   rcstool co /etc/fstab
2. Edit /etc/fstab
3. Add nfsvers=3 to line related to PDU share.
   Line will typically look like:
   
   x.y.z.t:/pdu_0 /opt/TKLCixp/pdu/mnt_n01 nfs rw,rsize=32768,wsize=32768,soft,nfsvers=3 0 0
4. Save and exit
5. Check back in:
   
   rcstool ci /etc/fstab
6. Reboot server

Dataexport

Following workaround must be applied on all base servers.

1. Stop dataexport process:
   
   pm.set off DataExport
2. Wait for process to stop
3. Edit /usr/TKLC/TKLCdataexport/prod/resource/dataexport.properties
4. Add "-o nfsvers=3" to line "dataexport.transport.nfs.mount.option"
   Line will look like:
   
   dataexport.transport.nfs.mount.option=-o soft -o nfsvers=3
5. Start dataexport process:
6. pm.set on DataExport

This solution has no attachment