Re: systems hangs every few days
On 2013-06-18 14:03, Jape Person wrote:
> On 06/18/2013 10:31 AM, Chris Purves wrote:
>> On 2013-06-18 10:20, Rob Owens wrote:
>>> ----- Original Message -----
>>>> From: "Chris Purves" <chris@northfolk.ca>
>>>> To: debian-user@lists.debian.org
>>>> Sent: Tuesday, June 18, 2013 8:59:07 AM
>>>> Subject: systems hangs every few days
>>>>
>>>> After upgrading to wheezy, I get a system hang every one or two days
>>>> where the system becomes completely unresponsive and I need do a
>>>> cold boot.
>>>>
>>> That sounds a lot like bad RAM. Run MEMTEST86 on it. I'm not sure if the debian install cds contain that, but Knoppix usually has that as a boot option. The full test could take more than a day. If you're lucky, it'll find an error in the first 10 minutes.
>>>
>>
>> I ran memtest86 already for about 20 minutes without error. (I forgot to mention that in original post). I will run it overnight tonight to see if anything else pops up.
>>
>> I am skeptical that it's bad RAM since the problem occurred after upgrading and I would have expected some kernel panic errors in the logs or something like that, although I suppose bad RAM can manifest itself in different ways.
>
> One thing I noted in your first message, Chris, was that it usually happens when
> a cron job is running. I immediately thought of two possibilities:
>
> 1. One or more of the cron jobs are running a process that is "tickling" a
> driver that doesn't work as well with the newer kernel as it did with the
> previous one.
>
> 2. A scheduled job is doing something CPU-intensive that could be making the
> system overheat. (I suppose load caused by a given job might be worse under
> wheezy than it was under squeeze.)
>
I looked more closely at the syslog just before the system hangs, and not always, but often the last line is a cron job calling a script that I wrote which uses phantomjs to log into the web GUI of my router to see if the IP address has been changed. I started running the script in a loop and twice after about 300 iterations the system would freeze up. I removed that script and have since reached an uptime of 8 days.
Interestingly, phantomjs is one of two packages that couldn't upgrade to wheezy. I'm still using the squeeze install. I have no idea how that could be the cause.
I'll report back in a few weeks to confirm if things are still running properly.
--
Chris Purves
Visit my blog: http://chris.northfolk.ca
"I can calculate the motion of heavenly bodies, but not the madness of people." - Sir Isaac Newton
Reply to: