Gabriel George POPA wrote:
I have two small questions:
1) When the OS generates too much messages, old messages are lost
(oldest lines present in `dmesg` are lost).
`dmesg' displays the system message buffer, which has a limited space.
Therefore, when it is full, it starts overwriting itself. Thus, lost
messages are indeed lost.
What can I do to see ALL messages ever recorded for dmesg printing?
You could write some script that `dmesg > /some/where/$time`
periodically (cron job).
More precisely, take a look at my `dmesg`:
# dmesg
arp info overwritten for 193.231.39.129 by 00:90:bf:10:88:40 on vr0
arp info overwritten for 193.231.39.129 by 00:10:dc:4c:6f:6c on vr0
arp info overwritten for 193.231.39.36 by 00:15:f2:16:f8:b4 on vr0
arp info overwritten for 193.231.39.54 by 00:e0:29:9b:bc:6c on vr0
...
(and a lot of other similar messages, similar if not even identical)
Most questions on this mail list require me to provide a valid output of
dmesg. But if old messages are erased, how am I
supposed to do this?
$ cat /var/run/dmesg.boot
I am not allowed to reboot the machine! The machine
is supposed to be running 24/7, NO reboot allowed.
Isn't this machine ever upgraded? Or, if it is so important - what about
redundancy? Well well.
2) What do these lines mean (the lines I copied above from the output of
`dmesg`)?
I'd say some machines are fighting over the same ip address. I could be
wrong, though. Don know, but I get a feeling that some failover
solution(s) could cause this.
/Alexander
