blob: e0513da17cf36c695368b008f1bb998984c38732 [file] [log] [blame]
Changes since V3.0.1
hardaker 21 Sep 95 09:27:18
(snmplib/md5.c) Forgot to include sys/types.h for u_long definition.
-------------------------------------------------------------------------------
Changes since V3.0.1
hardaker 26 Jul 95 16:31:41
- Added a new #define: EXIT_ON_BAD_KLREAD to config.h
If defined, and an error occurs while reading the kernel contents,
the agent will exit. This was the default. I wanted it to
continue.
- Port to dec-alpha/OSF 3.1
- alpha's long is 8 bytes, not 4.
- should be (mostly) 64 bit clean now.
- cleaned up a LOT of bugs CMU code in the process.
hardaker 28 Jul 95 14:23:41
Yet more problems fixed in the alpha port.
-- more int -> long checking.
hardaker 31 Jul 95 16:27:47
- snmpcheck changes:
- terminal (ascii) based supported fixed.
- Detects presence of a DISPLAY variable.
- Now requires the Term::Readkey package.
- asks users weather to fix problems
- raises and deiconifies when new problems are found
- sets the background color of new problems to pinkish
- new preference: ping host?
- global key bindings:
ctrl-q = quit; ctrl-h = hide; ctrl-s = seen all
hardaker 1 Sep 95 11:57:52
- Fixed alpha's kernel lookups:
* CMU code was passing ints into lseek (needs long).
hardaker 18 Sep 95 16:09:25
Minor mods to README file.
hardaker 18 Sep 95 16:13:49
Minor bug:
(sh_count_procs) check if returned file descriptor > 0 before continuing
-------------------------------------------------------------------------------
Changes since V3.0
hardaker 16 Jun 95 10:13:53
Removed 'gmake' -> 'make' in agent/Makefile
hardaker 16 Jun 95 10:15:42
House cleaning
- Compiles cleanly on hpux/gcc
(except one include file error I don't have control over.)
- Change VERSION_ID -> VERSIONID to avoid conflicts with HP's includes.
hardaker 16 Jun 95 14:13:53
House Cleaning:
Took out gcc/ansi requirement for the agent.
perror() is called more frequently for log archival
New Bug/Haven't solved:
HP-UX: if agent/*.c and agent/extensible/*.c are built with gcc,
DisplayStrings return garbage. No clue as to why.
hardaker 16 Jun 95 14:20:48
House Cleaning:
Missed a VERSION_ID
hardaker 21 Jun 95 09:11:27
Changed another "empty" error message in the CMU code to a real warning.
hardaker 21 Jun 95 09:15:31
Bug Fix:
Stopped returning a pointer to a non-static char array... (-> static char)
Passing wrong pointer to seterrorstatus() in setPerrorstatus()
Feature:
Added a setable mib entry to the version mib to re-read the config file(s).
hardaker 21 Jun 95 09:18:37
ece-mib.def:
Bug Fix:
Loadave mib: Integer -> DisplayString
versionClearCache: -> read-write (was read-only)
Feature:
Added a setable mib entry to the version mib to re-read the config file(s).
hardaker 21 Jun 95 09:19:17
Changed default CC to cc again.
hardaker 21 Jun 95 09:20:23
snmplib:
Bug fix: fdset -> (int *) fdset in select call.
hardaker 10 Jul 95 11:27:00
Two new command line options:
-l LOGFILE Outputs stderr/out to LOGFILE (overrides config.h def)
-L Do NOT write to a log file (overrides config.h)
hardaker 10 Jul 95 12:29:32
- More perror()s converted to set the snmp error status flag on the agent.
- error status flag is now prioritized. perrors get the highest so
other errors won't override them if they happen at a later time.
- Added the beginnings of a restart mib to the version mib of the
agent. Setting this writable mib to one will cause it to exec
itself again (for remote restart/updates).
hardaker 10 Jul 95 12:30:19
Minor comment change in config.h
hardaker 14 Jul 95 15:12:40
- Finished the restart mib. This is a setable sub-mib of the version
mib, default to VERSION.12(.1). If snmp-set to '1', it will restart
the agent.
* A side effect of this is that it restarts 3 times or so, since it
never returns a responce to the program generating the request.
If said program tries to resend the request (which most do), it
will AGAIN restart the agent... oh well.
hardaker 14 Jul 95 15:15:31
- snmpcheck (tk-perl-b4 script) is pretty stable at this point. I
lost curses support, and need to re-add it still, so it is still
beta. It does expect a log file of problems to exist (point it to
/dev/null), and needs to know the locations of the mibs to check
(which aren't configured by 'make'), so it needs some work.
However, it will query an agent, check its problems based on the
status of the error flag (or return code) being >1 and report them
with neat GUI-buttons and stuff (oh wow). No documentation on it though.
-------------------------------------------------------------------------------
Changes since 3-0A1:
hardaker 2 Jun 95 16:23:46
- version update
hardaker 5 Jun 95 12:44:46
- agent/extensible/Makefile:
-- no longer requires GNU make
-- etags no longer run by default
- SunOS additions:
-- loadaverage checking MIB enabled
-- disk checking MIB enabled
hardaker 9 Jun 95 10:21:07
- Added a SNMPBLOCK option to the config.h file. Supposedily setting
this to zero (now the default) will allow snmp requests to time out.
Haven't tested it throughly.
-- Contributed by Sanjai Narain <narain@thumper.bellcore.com>
hardaker 9 Jun 95 10:21:51
- Correctly ordered the mib table list.
-- Need to write a dynamic checker/re-orderer.
hardaker 9 Jun 95 10:23:35
- Correct usage for snmpwalk is now possible. IE, snmpwalk target
.1.3.6.1 will now browse the entire tree.
-- Still fails on relocatable extensible mibs and will until I
write a dynamic tree builder.
hardaker 9 Jun 95 14:08:15
- Changed test -x to test -f in snmplib/Makefile. Not all OSes
(ie ultrix) handle this.
hardaker 12 Jun 95 10:35:38
- local/trapnotif:
-- uses snmpv2 now (correctly)
hardaker 12 Jun 95 10:37:46
- Reports all errors through the error mib now instead of through perror().
- Cleaned up error handling. Shouldn't handle errors nicely if system
errors occur when scanning an extensible mib.
hardaker 14 Jun 95 11:24:20
agent:
- Mib table is now qsorted at run time and the extensible variables are
sorted in at this time. This brings the agent up to correct
SNMPwalk/getnext protocols. You can now walk the entire tree
including the extensible sections and it won't stop.
* This has an added bonus of checking the mib table
(agent/snmp_vars.c) for correct order. I have added things to the
wrong location too many times and caused the agent to function
incorrectly. Major win.
hardaker 14 Jun 95 12:56:49
- Cleaned up/Added more documentation to config.h
- GLOBALSECURITY now uses the left most four bits in the MIB acl.
This is more compatible with the CMU code than using the left most
2, which was done previously.
hardaker 14 Jun 95 13:43:38
- Added more documentation to the README file.
- Created a mailing list to handle announcements
(ucd-snmp-announce@ece.ucdavis.edu)