Maximum responsePdu packet size

2004-06-19 Thread Barb Roesch
What is the maximum packet size that the snmpd will send? I found the macro defines #define SNMP_MAX_MSG_SIZE 1472 #define SNMP_MAX_PDU_SIZE 64000 I'm assuming the SNMP_MAX_PDU_SIZE must be the upper threshold since my getBulk requests will return a snmp payload size larger than 1472. --

Re: Host Resource MIB on Broken Linux

2004-06-19 Thread Andy Smith
The #ifdeffed code may be needed on other platforms as well as Solaris. That was why I recommended that particular line of code. I only have access to Windows and Linux, but I would suggest testing on other platforms only excluding Linux from the ifdef. Andy Mike Slifcak wrote: The attached pat

Re: Outstanding Solaris bugs - can we fix for 5.1.2?

2004-06-19 Thread Mike Slifcak
Robert Story (Coders) wrote: On Thu, 17 Jun 2004 22:39:13 -0400 Mike wrote: MS> > 1.[ 956272 ] Solaris hostres hrPartitionSize values 8 times too big MS> > MS> >>This is the biggie. It's reporting incorrect values and the fix is MS> >>relatively simple. MS> MS> patch has been applied to V5-1-pat

Re: Host Resource MIB on Broken Linux

2004-06-19 Thread Mike Slifcak
The attached patch was added to patch 956272 which has the desired effect. #if defined(STRUCT_STATVFS_HAS_F_FRSIZE) && !defined(linux) Andy Smith wrote: Try #if defined(STRUCT_STATVFS_HAS_F_FRSIZE) && !defined(linux) That worked for me. Andy Bruce Shaw wrote: ARGG!! Stupid newbie question. H

Re: pass, exec, possibly more broken

2004-06-19 Thread Mike Slifcak
Robert Story (Coders) wrote: On Thu, 17 Jun 2004 23:21:40 -0400 Mike wrote: MS> Mike Slifcak wrote: MS> > William A.Carrel wrote: MS> > MS> >> In agent/mibgroup/util_funcs.c:~228, cachebytes needs to be ssize_t. MS> >> Left as size_t breaks the if(cachebytes < 0) when the read() above it MS> >

Blocking handler

2004-06-19 Thread Rodolfo Giometti
Hi, can an handler be blocked during a get/set by a read()/write() system call? For instance, if in the MODE_GET case I use a read() syscall to get a variable value and it blocks the the whole agent blocks too? If so, which would be the correct solution? Thanks, Rodolfo -- GNU/Linux Solutio