[AGENT++] Proposal: Explicit save of persistent objects via SNMP

Meena Nallasivan meena.nallasivan at masconit.com
Fri Oct 29 14:53:18 CEST 2004


Hi,

we are facing a memory leak in agentpp. Please find the trace attached
at the end of this mail.

Any help appreciated.


TRACE from the core
----------------------


#4  0x401c4d16 in __cxa_throw () from /usr/lib/libstdc++.so.5
#5  0x401c4f02 in operator new(unsigned) () from /usr/lib/libstdc++.so.5
#6  0x401c4fef in operator new[](unsigned) () from
/usr/lib/libstdc++.so.5
#7  0x086b17c1 in Oid::operator=(Oid const&) (this=0x411da8a8,
oid=@0x1a266028) at oid.cpp:202
#8  0x086b6525 in Vb::get_oid(Oid&) const (this=0x1a266028,
oid=@0x411da8a8) at vb.h:149
#9  0x086bccdf in Vb::operator=(Vb const&) (this=0x411da8a8,
vb=@0x1a266028) at vb.cpp:101
#10 0x086b374e in Pdu::get_vb(Vb&, int) const (this=0x1a273930,
vb=@0x411da8a8, index=0) at pdu.cpp:257
#11 0x083a0ee7 in Agentpp::Request::get_oid(int) (this=0x1a273160, i=0)
at request.cpp:511
#12 0x083944cb in Agentpp::Mib::process_request(Agentpp::Request*, int)
(this=0xc98ca00, req=0x1a273160,
    reqind=0) at mib.cpp:3201
#13 0x08394fcb in Agentpp::Mib::do_process_request(Agentpp::Request*)
(this=0xc98ca00, req=0x1a273160)
    at mib.cpp:3459
#14 0x0839ae53 in Agentpp::MibTask::run() (this=0x1a250330) at
threads.cpp:826
#15 0x0839a399 in Agentpp::TaskManager::run() (this=0xc98d1f8) at
threads.cpp:738
#






Meena





On Fri, 2004-10-22 at 12:32, Frank Fock wrote:
> I will add your comments to the MIB and also the autoSaveOnExit
> column. Thanks a lot for your feedback!
> 
> Best regards,
> Frank
> 
> Henning Eggers wrote:
> 
> >>If there are no objections posted on the mailing list, I will start
> >>development of the MIB's implementation next week. Neverthelles,
> >>comments and suggestions are still welcome!
> >>    
> >>
> >
> >This looks fine to me, I like the table-based approach a lot. Just a few
> >little things:
> >
> >The purpose of the permenant entry and the others should be made clearer in
> >the description for the table. So it should be extended to read (changes are
> >*marked*):
> >
> >"...  used to load the agent's configuration
> >*at boot time*. This entry will have a storage type of
> >'permanent' and thus it cannot be deleted.
> >*Other entries may be added to provide for
> >experimental or backup storage.*"
> >
> >
> >I also think that changing StoragePath would result in the timestamps
> >becoming invalid because they refer to the old path. So this behavior should
> >be described as follows (added to the descriptions of both timestamps):
> >"Will be reset to 0 if agentppCfgStoragePath is altered for this row."
> >The StoragePath of the default row cannot be changable, though, because that
> >would leave us with the classic hen-egg dilemma...
> >
> >And one more:
> >What about an "AutosaveOnExit" flag to control the current behavior of
> >saving the current configuration when the agents exits? Setting it for a row
> >would result in this storage being written on exit, leaving multiple copies
> >of the configuration. But I am not sure if this is really useful.
> >
> >Thanks for the work.
> >
> >Regards,
> >Henning
> >
> >_______________________________________________
> >AGENTPP mailing list
> >AGENTPP at agentpp.org
> >http://agentpp.org/mailman/listinfo/agentpp
> >
> >  
> >
> 
> 
> _______________________________________________
> AGENTPP mailing list
> AGENTPP at agentpp.org
> http://agentpp.org/mailman/listinfo/agentpp



More information about the AGENTPP mailing list