[Olsr-dev] [OLSRd 0000047]: memory leak on one remote end of point to point link

[(spam-protected)]

aljmas-luka-link-02 is client that is connected to aljmas-luka-link-01 (AP).
there is no issue on eth0 ports, they are stable, so there is probably
no issue with ethernet cable, only with wifi...

(spam-protected):~# logread |grep down
Mon Nov 17 00:04:19 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 00:14:06 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 00:17:00 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 00:29:39 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 00:42:17 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 00:55:10 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 01:47:32 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 01:55:48 2014 daemon.notice netifd: Network device 'wlan0'
link is down
Mon Nov 17 02:13:01 2014 daemon.notice netifd: Network device 'wlan0'
link is down

and from AP site here is signal level:

(spam-protected):~# iw dev wlan0 station dump
Station 00:27:22:dc:91:b2 (on wlan0)
        inactive time:  210 ms
        rx bytes:       3651834
        rx packets:     8080
        tx bytes:       391875
        tx packets:     2736
        tx retries:     10
        tx failed:      0
        signal:         -54 [-54] dBm
        signal avg:     -54 [-54] dBm
        tx bitrate:     58.5 MBit/s MCS 6
        rx bitrate:     65.0 MBit/s MCS 7
        authorized:     yes
        authenticated:  yes
        preamble:       long
        WMM/WME:        yes
        MFP:            no
        TDLS peer:      no

On Mon, Nov 24, 2014 at 2:59 PM, Henning Rogge <(spam-protected)> wrote:
> No problem...
>
> maybe you could do it even less complicated... count the number of
> "interface up/down" events within 24 hours on both nodes (or maybe
> less than 24 hours).
>
> I would just be interested if there is a relevant difference.
>
> Henning Rogge
>
> On Mon, Nov 24, 2014 at 2:57 PM, (spam-protected)
> <(spam-protected)> wrote:
>> I'll check logs on all other nodes, after this problematic
>> aljmas-luka-link-02 node hotspot is connected via UTP cable, and
>> infront of aljmas-luka-link-02 there is wifi link towards
>> aljmas-luka-link-01.
>> I didn't see wifi breaking that ofter, so I'll concentrate my
>> investigation on lan cable (could be at fault) between
>> aljmas-luka-link-02 and aljmas-luka node.
>>
>> I'll report back with logs from all nodes... sorry for the delay.
>>
>> On Sat, Nov 15, 2014 at 9:48 AM, Henning Rogge <(spam-protected)> wrote:
>>> Hi,
>>>
>>> I just had a look at the syslog again and I think there are quite a
>>> few "interface added/removed" messages in this... is this normal on
>>> the other "not memory leeking" node too?
>>>
>>> Henning Rogge
>>>
>>> On Fri, Nov 14, 2014 at 10:30 AM, Jernej Kos <(spam-protected)> wrote:
>>>> Hello!
>>>>
>>>> On 14. 11. 2014 10:22, Henning Rogge wrote:
>>>>> just to be absolutely sure, you verified that its olsrd that consumes
>>>>> the storage, right?
>>>>
>>>> Valent did the checks, he said that the size of the virtual memory
>>>> allocated for the olsrd process is constantly rising and causes the node
>>>> to go out of memory if left unchecked.
>>>>
>>>> Note that we are only seeing this behaviour on this one node.
>>>>
>>>>> hmm, is there a computer with a little bit more storage attached to
>>>>> the node with the memory leak? If yes we could maybe take a longer
>>>>> binary tcpdump and replay it to an olsrd with a valgrind attached (I
>>>>> don't think we can run valgrind on your node).
>>>>
>>>> No, I think that valgrind would not fit on this node. As far as a
>>>> computer with more storage goes, Valent should be able to answer that.
>>>>
>>>>
>>>> Jernej
>>>>