date in binary and messages differ one day

[Michael Schulte zur Sur]

Hi all,

I have noticed that the date in the binary errorlog and the messages differ excatly one day for one speficic event. How can that be?
The os is 5.1B pk4.

Thanks for any idea,

Michael

[Stuart Fuller_2]

What is the specific event - can you post an example?

What error log reporting tool are you using (uerf, decevent, sea...)?

If the specific event is regarding an external controller, like an HSZ80, the date on the external controller could be wrong.

More date (umm, I mean, data) is needed.

[Michael Schulte zur Sur]

Stuart,
thanks for answering. I used decevent for analyzing. Both refer to a boot.

messages:
Jun 11 13:45:59 pra02d vmunix: physical memory = 1024.00 megabytes.
Jun 11 13:45:59 pra02d vmunix: available memory = 979.62 megabytes.
Jun 11 13:45:59 pra02d vmunix: using 1293 buffers containing 10.10 megabytes of
memory
Jun 11 13:45:59 pra02d vmunix: Firmware revision: 6.9-1

errorlog:
Logging OS 2. Digital UNIX
System Architecture 2. Alpha
Event sequence number 1.
Timestamp of occurrence 10-JUN-2005 13:46:00
Host name pra02d

System type register x00000022 Systype 34. (Regatta Family)
Number of CPUs (mpnum) x00000001
CPU logging event (mperr) x00000000

Event validity 1. O/S claims event is valid
Event severity 5. Low Priority
Entry type 300. Start-Up ASCII Message Type
-1. - (minor class)

SWI Minor class 9. ASCII Message
SWI Minor sub class 3. Startup

ASCII Message
Memory trolling not supported, cpu Major id 8, Minor id 4
Alpha boot: available memory from 0x2b9e000 to 0x3ff40000
Compaq Tru64 UNIX V5.1B (Rev. 2650); Tue Mar 22 15:38:17 CET 2005
physical memory = 1024.00 megabytes.
available memory = 979.62 megabytes.

[Michael Schulte zur Sur]

The correct date is Jun 11th because on that Saturday I booted the machine after a planned outage of a few hours.

Michael

[Stuart Fuller_2]

That's a weird one, alright.

The boot message in the error log shows 10-Jun, and this is the system date at the time of the boot.

The syslog messages shows 11-Jun.

Therefore, the system time changed between boot time and when syslog started logging messages.

The default order of startup is syslogd, binlogd, some network stuff, then the time setter and the time daemon.

Is it possible that the order of startups have changed, such that syslog starts after the time setting/daemon? I suppose that it's also possible that syslogd started up, and took a little while to actually "get going", and that the time setter ran during that interval.

Of course, that doesn't explain why the system time at boot was 10-Jun.

When was the shutdown? Do the error log and syslog messages indicate anything out of the ordinary?

[Michael Schulte zur Sur]

The shutdown was a planned one and occurred round 6am on Jun 11th. There were no out of the ordinary messages at that time. The sequence of events, you mentioned, is the same. So I still do not have a clue!

Thanks,

Michael

[Ronald_35]

Did you finally get a fix for this one ?

[Michael Schulte zur Sur]

Ronald,

thanks for posting but the problem is a mystery to us.

greetings,

Michael

[Han Pilmeyer · Accepted solution]

This is a known problem with DECevent. It displays dates as one day early. Any other utility will display the correct date. I would recommend using WEBES.

[Michael Schulte zur Sur]

Han,
I can't believe this has eluded me so long.

Ronald,
thanks to you this thread got once more attention.

Michael