Server drops to OK prompt everynight?

[mranderson22]

When I come into work our UNIX server has dropped to the OK prompt and cannot be seen on the network until I type "go". This has only started since the server crashed last week any suggestions?

 

[dbase77]

hi,

can you check if you have set "auto-boot" to false? you can check from command line as root. type "eeprom" should give you a list of options. make sure it set to true for auto-boot, if not it will stop at OK prompt.

regards,
feroz

 

[tuka]

When you type go , did the system come back to normal operations?

If so than a crash probably didn't occur. Is a Sun's keyboard connected? If a keyboard is connected there is a possibilty the keyboard is intermittent or getting disconnected.

Run the command eeprom and look at the auto-boot? parameter.
If auto-boot? is set to true than if the system crashes a reboot would take place. This can be seen in the /var/adm/messages file.


1. A power glitch (loss) would take the system down and if the auto-boot: was set to false then the system would go to the ok prompt BUT you couldn't type "go" and CONTINUE.

2. A hardware/software problem normally create a panic which can be seen in the /var/adm/messages file and a reboot would be attempted. Typing "go" would not help.


What is your server's architecture?

 

[mranderson22]

The server is SunFire V100 it is a rack server and has no keyboard or mouse connected to it it does work fine once you type go so I dont think it crashes overnight I do a backup of files overnight which writes to another server I think once it has finished this it drops to the ok prompt for some reason. I will look at the auto-boot setting now.

 

[julianbarnett]

I'm assuming powerd isn't running - it shouldn't be since you've had no problems in the past. Check /var/adm/messages for a panic - could well be a faulty memory problem - I've seen this cause repeated powerdowns. Also check your system panic dumps are being saved and processed - call dumpadm to see the current settings. Saved core files from a system panic are by default saved in /var/crash/systemname and may be required by Sun support to examine hardware faults.

Finally check your crontabs - make sure no ones added an init, reboot or shutdown command.

 

[crowe]

Are there any serial connections to the server? It almost sounds like someone (or something) is sending a break sequence, causing the system to drop down to the ok prompt.

crowe

 

[wtrepani]

If there is a serial terminal attached to it, that could cause what you are seeing. If a serial terminal is attached at boot, then it is considered the console. If it is disconnected, powered off, or the connection is broken in any way, it will stop the system and drop you to an OK prompt. I think there is a eeprom setting to stop this, but I don't know what it is off the top of my head...

Will

 

[SamBones]

Type "last reboot" to see when they've been happening. If it's a consistent time, then it very well could be a cron job. If it's all over the place, it could be hardware, software, human error, or just about anything.

 

[tarn]

What does your backup script do?

If its like (some that I've written) it takes the server into init S for a level 0 then at the end pops it back into init 3 "has someone tampered with your backup script? have you commented something out in it or choped it around lately ?).

Also where does this backup write to ? its not filling root or trying to write to a "broken" partition that is unstable since the "Crash"?

Good Luck

Laurie.