dsm_sa_snmpd is leaking on the order of 20 Gigabytes per day on a number
of Ubuntu 10.04 64-bit systems.  I'm seeing this on R905s as well as
2950s.  The version of Dell OMSA is 6.0.1-10 and is from the Dell
provided Ubuntu package.

The only thing querying Dell SNMP MIB data is the mtk-nagios plugin:
http://www.hpccommunity.org/downloads.php?do=file&id=101

I do not see this problem on 32-bit 10.04 running Dell OMSA 6.0.1-10
packaged by SARA.  (AFAICT Dell has not yet published a 32-bit Ubuntu
package.)

I do not see this problem on 64-bit RedHat running Dell OMSA 5.5.0
(packaged by Dell.)  Though on this system the 32-bit dsm_sa_snmpd is
running.

Looking through the archives and other sources, I see reports of memory
leaks, but no resolution:

http://lists.us.dell.com/pipermail/linux-poweredge/2007-February/029648.html
http://lists.us.dell.com/pipermail/linux-poweredge/2007-March/029972.html
https://subtrac.sara.nl/oss/omsa_2_deb/ticket/66

I've been strace-ing the processes on the leaking and non-leaking
systems.  On the leaking systems, the process leaks 94MB over the course
of 9194 brk() calls.  The pattern is mostly: allocate 1MB, allocate 1MB,
free 2MB.  However, there are the following leaks:

 Count   Size (Bytes)
     11 4096
      1 8192
     66 12288
    385 16384
      4 20480
      1 24576
      1 28672
      1 32768
      2 65536
      1 69632
      1 81920
      2 90112
      2 143360
      1 147456
      3 1032192

I also see the process calling madvise() on large ranges of memory after
it has grown to several GB in size.

I don't believe there is source code available.  If there is, please
point me at it.  Otherwise, hopefully the data above will help Dell
track down the source of this memory leak.

Suggestions and solutions would be very welcome.

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge@dell.com
https://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq

Reply via email to