[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Weirdness with init

Steve wrote:

Just wondering if anyone has ever seen something like this before. I have a Debian Etch system running a custom kernel based off of stock kernels from kernel.org. The system was upgraded from Sarge to Etch and because of the custom kernel wasn't rebooted. That upgrade took place months ago and every was running smoothly.

Today I noticed when running the 'uptime' command that the uptime was listed as 2 days 6 hours and some change. I was expecting to see something close to 500 days. I investigated all of the logs and could find absolutely no evidence of a reboot. The system is fairly heavily monitored so I would've seen at least one or two e-mails from the reboot. I also have no reason to believe, based on the initial investigation, that there is anything malicious involved.

Running ps, it looked like init and related processes all had a timestamp of Jan 4, two days ago. A few other processes had a timestamp of just '2009' while most had a 'correct' (or sane) timestamp.

I'm trying to think of logical reasons for this to have happened. Was there possibly an update that did something to init or related? Could there have been a problem with the hardware clock on the server temporarily? Could something have gone wrong with ntpd that asjusted the time and then corrected itself? Was the uptime just sufficiently long enough that something rolled over? (It's a fairly heavily used server.)

Anyone else seen something similar or have other thoughts?

Older kernels have a bug where the uptime rolls over after about 497 days (2^32 hundreds of a second). This would explain the low uptime your server reports. I currently have no access to such a system, so I don't know if this also can explain the other symptoms you're seeing.

Reply to: