> What about the information from top? When comparing RHEL5 and RHEL6 systems, > I would compare the total CPU usage of the server (out of 100% not 2400% or > 1600%). > > Since the hardware is different, comparing a 16 named threads on a 16 core > box at ???MHz against a 24 core box with 24 named threads at ???MHz may not > necessarily be valid. If the CPUs are running at the same frequency (look at > what speed they are actually running at vs the max speed... see > /proc/cpuinfo ) then you can probably account for the 16 vs 24 core > difference pretty easily. If the CPUs run at more than negligibly different > frequencies you will have to factor that into any comparison or make the > frequencies the same to make a 1:1 good comparison.
The systems run at the exact same frequency processors (RHEL 6 - X5675 @ 3.07GHz, RHEL 5 - X5667 @ 3.07GHz). One is just a little older and only has 16 CPUs. I've run named on the RHEL 6 system with only 16 procs (named -n 16) to see if it made a difference and the result was the same, named on the RHEL 6 system running 6-7 times the load of a RHEL 5 system. We aren't running DNSSec so I don't think the managed-keys-directory should be an issue. Running a 30 second strace on one of the named threads shows the process is a lot busier and has more errors on the RHEL 6 system: RHEL 6: > strace -c -p 29904 Process 29904 attached - interrupt to quit Process 29904 detached % time seconds usecs/call calls errors syscall ------ ----------- ----------- --------- --------- ---------------- 99.93 9.074464 135 67128 20375 futex 0.04 0.003689 1 2538 2430 recvmsg 0.02 0.002141 1 2498 write 0.00 0.000256 2 103 sendmsg 0.00 0.000138 4 36 socket 0.00 0.000065 2 36 connect 0.00 0.000019 0 72 setsockopt 0.00 0.000000 0 36 close 0.00 0.000000 0 36 bind 0.00 0.000000 0 36 getsockopt 0.00 0.000000 0 108 fcntl ------ ----------- ----------- --------- --------- ---------------- 100.00 9.080772 72627 22805 total RHEL 5: > strace -c -p 18498 Process 18498 attached - interrupt to quit Process 18498 detached % time seconds usecs/call calls errors syscall ------ ----------- ----------- --------- --------- ---------------- 99.97 1.549134 69 22399 5604 futex 0.02 0.000243 0 720 698 recvmsg 0.01 0.000193 0 722 write 0.00 0.000017 0 36 socket 0.00 0.000000 0 36 close 0.00 0.000000 0 36 connect 0.00 0.000000 0 39 sendmsg 0.00 0.000000 0 36 bind 0.00 0.000000 0 72 setsockopt 0.00 0.000000 0 36 getsockopt 0.00 0.000000 0 108 fcntl ------ ----------- ----------- --------- --------- ---------------- 100.00 1.549587 24240 6302 total Are others seeing the named process run at 130-180% on RHEL 6? We've never seen this high of CPU usage for named on any system including a Solaris 10 system running 32 CPUs. They have all run around 11-30% CPU depending on the time of day. -- Daniel _______________________________________________ Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users
