Forum Discussion

rraja's avatar
rraja
Level 4
13 years ago

bpdbm heavy CPU usage - a lot of time in kernel after upgrading to 7.1.0.2

After upgrading to 7.1.0.2, constant 100% or near 100% CPU busy.

 

The EEBs that we installed on the master are:

eebinstaller.2529202.1.solaris fixes a problem with incremental backup expired early. (This one has the bpdbm binary.)

eebinstaller.2535076.1.solaris is the NDMP bundle with other fixes. 

We didn’t run v7.1.0.2 without the EEBs.  We installed the EEBs immediately after upgrading from v7.0.1 (with a few EEBs) to v7.1 then v7.1.0.2 then adding two EEBs. 

There isn’t a CPU spike as such – just a constant 100% or near 100% CPU busy. 

 

load averages: 10.18, 11.91, 12.38                                                                  10:37:22

182 processes: 172 sleeping, 6 running, 4 on cpu

CPU states:     % idle,     % user,     % kernel,     % iowait,     % swap

Memory: 16G real, 867M free, 2763M swap in use, 15G swap free

 

   PID USERNAME LWP PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND

 23353 root       1  22    0    0K    0K run      0:20  9.15% bpdbm

 10528 root       1  22    0    0K    0K run      6:38  8.40% bpdbm

 23852 root       1  31    0    0K    0K run      0:08  5.83% bpdbm

 23719 root       1  32    0    0K    0K run      0:07  4.66% bpdbm

 23815 root       1  32    0    0K    0K run      0:04  3.39% bpdbm

 24083 root       1  22    0    0K    0K cpu/2    0:02  2.40% bpdbm

   795 root      44  59    0    0K    0K sleep  162.1H  2.35% dbsrv11

 21310 root       1  60    0    0K    0K run      0:13  1.76% bpdbm

   942 root      10  59    0    0K    0K sleep   54.9H  1.75% nbrb

 21313 root       1  48    0    0K    0K sleep    0:14  1.75% bpdbm

   939 root      58  59    0    0K    0K sleep  110.1H  1.59% nbemm

 21457 root       1  59    0    0K    0K sleep    0:11  1.51% bpdbm

 21390 root       1  59    0    0K    0K sleep    0:11  1.50% bpdbm

 21393 root       1  49    0    0K    0K sleep    0:11  1.48% bpdbm

 21423 root       1  54    0    0K    0K sleep    0:11  1.48% bpdbm

 

FYI., we have case opened with Symantec, they recommended us to restart the master server (we are not happy with this)

Please advice someone faced with similar issue / how to troubleshoot this.

5 Replies

  • Hi rraja,

     

    Do you have bpdbm logging enabled, with VERBOSE set to high? This may give some answers as to what is going on with bpdbm.

     

  • What backup jobs are running during this time?  I know mis-configured oracle backups can cause bpdbm to spike as it runs long searches to determine backup ID is valid.  %t missing for example can cause this:

    Put %t at the end of the backup file name format. NetBackup uses the timestamp
    as part of its search criteria for catalog images. Without this timestamp,
    performance might degrade as the NetBackup catalog grows.

    A RMAN cross check can cause this as well.  Just some ideas but more details as to what is running would help here.

  • If you are using NetBackup with NBAC turned ON , then you will see this. Get in touch with Symantec Suport. The problem has been identified as a NBAC issue and fix is available.