Swap space issue

[Kavitha]

Hi All,

We have around 7 Linux servers. All these servers swap space getting filled while the Netbackup starts. Applications are running in the server. We thought this issue is due to the application and we brought the applications offline and tested the backup. but still swap gets full. Symantec says this is the issue with the OS so, we will not support and contact the OS vendor. Redhat says that this issue is with the Netbackup.

Finaly Symantec suggested us to change the Kernel settings.

NETBKUP KERNEL PARAMETER SETTINGS : kernel.sem = 300 1024 32 1024
CURRENT KERNEL PARAMETER SETTINGS : kernel.sem = 250 32000 32 128

This didnt help... we are still facing the same issue.

We have 11 GB free space in swap before backup starts.. but it gets reduced to zero whn backup is running.

No one can able to access the Apps...since app also need some space. Increasing the swap is not the option bcoz backup fills all available swap space.

Please help me to solve this issue. I know only bit abt unix.

Thanks in advance...

[rizwan84tx]

Please provide the Operating System information and NBU version info always while posting your issues, so that everyone here can help in this forum.

[Nicolai]

Are the 7 servers Netbackup media servers or clients ?.

What does "ps -aux | awk '{print $4"\t"$11}' | sort | uniq -c | awk '{print $2" "$1" "$3}' | sort -nr" say when backup is running ?

[Kavitha]

@riwan - I am sorry. I will provide all the details.

@Nicolai - We have stopped the backup for this server. We need to schedule the change to run the backup again for those servers.since we have to bring the applications offline. We will schedule it ASAP.

[J_H_Is_gone]

Suggestion to try.

Say you are doing multi streaming, so you get 1 stream for each file system.

And say you have 5 file systems on the server, if you allow all 5 streams to run at once it may be too much.

Try going into the master server properties/client attributes and add the server and set it to only allow 1 or 2 streams to run at once.  So now you would get 5 jobs, 2 running and 3 queued.  this could limit the amount of overhead Netbackup is using.

[Nicolai]

It may be a run away configuration where multistreaming is configured for a directory with millions of files or sub-directories. This mean a single backup spans hundred (even 1000's) bpbkar processes.

[Kavitha]

Server Details:

uname o/p:

Linux <Server name>  2.6.18-128.el5 #1 SMP Wed Dec 17 11:41:38 EST 2008 x86_64 x86_64 x86_64 GNU/Linux

H/W: ProLiant DL585 G1

Netbackup Version: 6.5.4

Policy Configuration:

3 schedules ( daily, Weekly and monthly Full )

Backup selection : All_Local_Drives

Check points: every 15 mins

Cross mount point: Enabled

Follow NFS: Not enabled.

Compression and Encryption: Not enabled.

Mutiple Date stream: Not enabled.

True image restore and move detection: Not enabled.

Please let me know if you need any other information.

@Nicolai:  We have check with both mutiple stream enabled and disabled. but no use we are getting the swap filled for both the situations. we have cross verified the bpbkar counts also.