UPDATE#2: binlogd _at_host

From: Mark S. Yamnicky <marco_at_nirvana.ES.bss.boeing.com>
Date: Fri, 27 Jul 2001 14:08:11 -0700 (PDT)

so I get a reply from Serge P. that says, "why don't you run binlog
on jane in debug mode?" Uh... I dunno... guess I was all focused on
the client. Thanks Serge for that shot of espresso.

On that note my new errors are:

logevent()
        trailer check: 0
        hdrp=0x14000a800 reclen=0xfffe trailer=0x14001a7fa: 0???
        corrupted errlog record length or missing trailer!
        attempting to log record anyway...
        corrupted header record length value
        
Has anyone come across this before? I checked the archives and didn't
find anything that was really applicable.

Thanks for your time
/mark

------------- Begin Forwarded Message -------------

What seems very simplistic and many others have working is humbling me.
I've read the summaries and I know this thing works... but I'm still
missing a _tiny_ piece of the puzzel.

Were running 4.0F Patch level 4

I'm aware of the /etc/services issue and have corrected the binlogd entry.

The host alpha030 is configured to send log data to host jane.
alpha030 starts binlogd in debug mode and confirms the remote host is jane.
Execute "logger -b test" to log a message to jane.
binlogd debugger indicates the message was sent to jane.

On jane using tcpdump:

jane# tcpdump dst port 706
12:11:09.126376 alpha030.binlog > jane.binlog: udp 208

so one would think the data is getting to jane... but jane writes nothing
to its local log file. I can use logger on jane and the log file does get
updated.

I went as far creating new /dev/kbinlog and /dev/klog devices via MAKEDEV.

Two additional items crossed my path.

1) Version 5 binlogd manpage states a /etc/binlog.auth file should exists.
I tried an auth file on 4.0 using short names and the FQDN. It made no
difference.

2) On a system whose log file is on the local disk, stop binlogd and remove
the errorlog. Start binlogd and log a message via logger -b. The timestamp
on the message logged is 1 hour off. For example if at 12:30 I log a message,
uerf shows the timestamp as 13:30. This only happens for the very first
message. Subsequent message timestamps are correct.

 
Any other thoughts on what I may be missing?

Thanks
/mark







->Admins,
->
->Probably an easy question:
->
->When starting the binary event logger binlogd in debug mode, it complains that
->it cannot open the file associated with the /etc/binlog.conf entry:
->
->*.* /path/to/file_at_host
->
->however change this to:
->
->*.* _at_host
->
->as the binlogd manpage indicates and binlogd is happy. Using "logger" and
->binlogd in debug mode confirms that data is being sent to "host". Great. So
->where on "host" is the data being logged?
->
->Thanks
->/mark
->
Received on Fri Jul 27 2001 - 21:09:35 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:42 NZDT