Solaris

this system go rebooted itself and when I check the last reboot command , the dates are shown without any sequence...

Any idea?


# last reboot -a
reboot system boot Mon Jan 25 19:33
reboot system down Mon Jan 25 19:25
reboot system boot Tue Nov 10 14:33
reboot system down Mon Jan 25 19:25
reboot system boot Tue Nov 10 14:33
reboot system down Tue Nov 10 14:31
reboot system boot Mon Oct 26 12:07

what's your problem here? the output or system rebooted by itself ? provide necessary log outputs eg messages during the time.
include h/w type, kernel patch level and OBP level.

this is a Netra440 and system rebooted automatically.

there is nothing in dmesg except of reboot messages which come in normal reboot.

BTW, can this be a DHCP issue?

we have a DHCP based network and some IP addresses were reserved for server, after migrating the DHCP server, those IP Addresses are no more excluded from lease range and there was an IP address conflicting message just prior to abnormal reboot.

what's your OBP level? prtconf -V (maybe the cause)
OR
The watchdog XIR may have been caused by a heavy period of network activity keeping the ce Ethernet interrupt on CPU for a long time.
This is a known problem. The workaround is to increase the ALOM sys_wdttimeout setting. The default value is 60 seconds, this value can be doubled to 120 seconds.

sc> setsc sys_wdttimeout 120

The command "last" reads the system accounting file "wtmpx". If this file is not maintained it can accumulate for years. I don't think your entries are in random order, they just lack the year.

Untested method to find the vintage of your wtmpx file on Solaris 10.
Will not work if your wtmpx file is much over one Gigabyte.
Will need lots of disc space in system temporary directory (about twice the size of your wtmpx file).

Code:

cat /var/adm/wtmpx | /usr/lib/acct/fwtmp | head

---------- Post updated at 18:10 ---------- Previous update was at 17:51 ----------

BTW. Duplicate IP address can crash the system.