[ntp:hackers] Re: out of memory with rc1!

John Pettitt jpp at cloudview.com
Thu Mar 16 06:41:03 UTC 2006


Danny Mayer wrote:
> John Pettitt wrote:
>   
>> Danny Mayer wrote:
>>     
>>> Reg Clemens wrote:
>>>   
>>>       
>>>>> Danny Mayer wrote:
>>>>>
>>>>>       
>>>>>           
>>>>>> Can you comment out your NMEA refclock and see if it stabilizes? You
>>>>>> shouldn't see it increasing memory by much after the first few minutes.
>>>>>>
>>>>>>         
>>>>>>             
>>>>> I cut ntp conf back to this
>>>>>
>>>>> driftfile /var/db/ntpd.drift
>>>>> server 192.168.3.254
>>>>>
>>>>> and it's still growing.
>>>>>
>>>>>       
>>>>>           
>>>> Thats a bit strange, my assumption would be that without the NMEA driver being 
>>>> configured
>>>> it shouldnt be able to grab memory.
>>>>
>>>> But here I took rc1, built it w/o NMEA, just ONCORE and SHM, and there is no 
>>>> problem, no
>>>> memory gain over a period of 12hours.
>>>>     
>>>>         
>>> Thanks for that data point. Does that mean that if built with NMEA then
>>> even if you don't configure an NMEA refclock there is a leak while if
>>> built without it there is no leak? Is NMEA the common thread here or is
>>> that just coincidence?
>>>   
>>>       
>> coincidence - I just built it with --disable-all-clocks and it's still
>> growing on my FreeBSD box - - I think FreeBSD is the common thread here.  
>>
>>     
>
> I don't think so. We're running it on ntp1 which is a FreeBSD machine:
> FreeBSD 5.2.1-RELEASE-p13
>
> and it's been rock solid stable. It hasn't changed it's memory usage
> since this morning.
>
> Danny
>   
That's really odd because I can make it fail on FreeBSD 5.3, 5.4 and 6.0

>From bugzilla
> Additional Comments From kardel at ntp.org (Frank Kardel)
> Submitted on 2006-03-15 15:31
> At my site it dug out an intermittent bug in libc(NetBSD-current). I am
> fixing that. After that I'll check for more leaks (if any).

I wonder if libc changed between 5.2.1 and 5.3 (I seem to recall it did)
- oh well time to try a few releases and see where the change happened
that triggers this

john







More information about the hackers mailing list