Skip to contentSkip to conversation

conntrackd crash after few seconds

DISCUSSION Nicolas RENAULT 5 participants 7 messages Jun 5, 2014 – Jun 16, 2014 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Hello,
thanks for the work !
and sorry for bad english.

informations :

debian wheezy 3.2.0-4-amd64 #1 SMP Debian 3.2.57-3+deb7u2 x86_64 GNU/Linux
packages

ii  conntrack                        1:1.2.1-1 amd64        Program to modify the conntrack tables
ii  conntrackd                       1:1.2.1-1 amd64        Connection tracking daemon
ii  libnetfilter-conntrack3:amd64    1.0.1-1 amd64

conntrackd.conf
--------------------------------------------------

Sync {
    Mode FTFW {
        DisableExternalCache On
    }
    UDP {
        IPv4_address 10.98.145.5
        IPv4_Destination_Address 10.98.145.6
        Port 3780
        Interface eth2
        SndSocketBuffer 1249280
        RcvSocketBuffer 1249280
        Checksum on
    }
}
General {
    Nice -20
    Scheduler {
        Type FIFO
        Priority 99
    }
    HashSize 32768
    HashLimit 131072
    LogFile on
    Syslog off
    LockFile /var/lock/conntrack.lock
    UNIX {
        Path /var/run/conntrackd.ctl
        Backlog 20
    }
    NetlinkBufferSize 2097152
    NetlinkBufferSizeMaxGrowth 8388608
    NetlinkOverrunResync On
    NetlinkEventsReliable Off
    PollSecs 15
    EventIterationLimit 100
    Filter From Userspace {
        Protocol Accept {
            TCP
        }
        Address Ignore {
            IPv4_address 127.0.0.1
        }
    }
}

----------------------------------------------------------

My problem :

when a start de deamon : /etc/init.d/conntrackd start i can see this line for ten secondes in "ps faux"

root 3764 0.0 0.0 14836 2484 ? S<s 16:53 0:00 /usr/sbin/conntrackd -d -C /etc/conntrackd/conntrackd.conf

and then disapear.

a try to run with (after deleting /var/lock/conntrack.lock )

conntrackd -C /etc/conntrackd/conntrackd.conf

then i have :

conntrackd: api.c:526: __build_query_exp: Assertion `ssh != ((void *)0)' failed. abandon

at about ten seconds.

any ideas ?

others informations , i use the same conf file without any problems on squeeze (2.6.32-5) with .

ii  conntrack                          1:0.9.14-2 Program to modify the conntrack tables
ii  conntrackd                         1:0.9.14-2 Connection tracking daemon
ii  libnetfilter-conntrack3            0.0.101-1

only a every two hours avg load peak activity.

regards

Signature · 1 line
Nicolas
#1 lore raw
Quoted 3 lines · Nicolas RENAULT

informations :

debian wheezy 3.2.0-4-amd64 #1 SMP Debian 3.2.57-3+deb7u2 x86_64

Hi.

We had crashes on debian wheezy which were solved by upgrade to

linux-image-3.13-0.bpo.1-amd64

from wheezy backports.

mk

A message that is not in the archive (sent off-list or not yet received): CAOkSjBifDQQNNQJbkdraYxNvStvK9CrWRNO7LtawpJJ5n=cTMg@mail.gmail.com

#3 lore raw
Quoted 9 lines · Arturo Borrero Gonzalez
Quoted 7 lines · Martin Kraus
Quoted 3 lines · Nicolas RENAULT

informations :

debian wheezy 3.2.0-4-amd64 #1 SMP Debian 3.2.57-3+deb7u2 x86_64

Hi.

Quoted 7 lines
Quoted 4 lines

We had crashes on debian wheezy which were solved by upgrade to

linux-image-3.13-0.bpo.1-amd64

Is this the same as the Debian bug #746464 [0]?

probably not. we hit a kernel panic and when I redirected kernel console to the serial port for another run it showed that there was a stalled cpu core running conntrackd process and everything was just dead.

mk

#4 lore raw
Quoted 14 lines · Martin Kraus
Quoted 13 lines · Arturo Borrero Gonzalez
Quoted 10 lines · Martin Kraus
Quoted 3 lines · Nicolas RENAULT

informations :

debian wheezy 3.2.0-4-amd64 #1 SMP Debian 3.2.57-3+deb7u2 x86_64

Hi.

We had crashes on debian wheezy which were solved by upgrade to

linux-image-3.13-0.bpo.1-amd64

Quoted 6 lines
Quoted 1 line

Is this the same as the Debian bug #746464 [0]?

probably not. we hit a kernel panic and when I redirected kernel console to the serial port for another run it showed that there was a stalled cpu core running conntrackd process and everything was just dead.

mk

Hi,

thanks for the reply.

some more informations :

I don't have kernel panic and so on , as i say in the first post it's just conntrackd that stop running after about 10 secondes. no message in syslog or messages or deamon.log or conntrackd.log....

I use conntrackd but i'm not a specialist so when i say that a try to run it with

conntrackd -C /etc/conntrackd/conntrackd.conf i'm right ?

on the firt post my conntrackd.conf is good or not ? perhaps i made a mistake on the buffer size or thinks like that ?

I will try to update to linux-image-3.13-0.bpo.1-amd64 but i hope that it will not broke others thinks :)

regards

Signature · 1 line
Nicolas
#5 lore raw
In reply to Martin Kraus
Quoted 14 lines · Martin Kraus
Quoted 13 lines · Arturo Borrero Gonzalez
Quoted 10 lines · Martin Kraus
Quoted 3 lines · Nicolas RENAULT

informations :

debian wheezy 3.2.0-4-amd64 #1 SMP Debian 3.2.57-3+deb7u2 x86_64

Hi.

We had crashes on debian wheezy which were solved by upgrade to

linux-image-3.13-0.bpo.1-amd64

Quoted 6 lines
Quoted 1 line

Is this the same as the Debian bug #746464 [0]?

probably not. we hit a kernel panic and when I redirected kernel console to the serial port for another run it showed that there was a stalled cpu core running conntrackd process and everything was just dead.

mk

Hi,

I find the solution , the proble come from conntrackd.conf it was made from the one on a squeeze and with the example bring with the package a made this one :

Sync {
    Mode FTFW {
        DisableExternalCache On
    }
    UDP {
        IPv4_address 10.98.145.5
        IPv4_Destination_Address 10.98.145.6
        Port 3780
        Interface eth2
        SndSocketBuffer 1249280
        RcvSocketBuffer 1249280
        Checksum on
    }
}
General {
    HashSize 8192
    HashLimit 65535
    Syslog on
    LockFile /var/lock/conntrackd.lock
    UNIX {
        Path /var/run/conntrackd.sock
        Backlog 20
    }
    SocketBufferSize 262142
    SocketBufferSizeMaxGrown 655355
    Filter {
        Protocol Accept {
            TCP
        }
        Address Ignore {
            IPv4_address 10.98.0.0/16     #dedicated link #
        }
    }
}

diff :
old new
HashSize 32768 8192
HashLimit 131072 65535

- NetlinkBufferSize 2097152
- NetlinkBufferSizeMaxGrowth 8388608
- NetlinkOverrunResync On
- NetlinkEventsReliable Off
- PollSecs 15
- EventIterationLimit 100
- Filter From Userspace {

+ SocketBufferSize 262142
+ SocketBufferSizeMaxGrown 655355
+ Filter {

can someone look at my new conntrackd.conf and says if i forget somethinks important ?.

regards

Signature · 1 line
Nicolas
#6 lore raw
In reply to Nicolas RENAULT
Quoted 10 lines · Nicolas RENAULT

others informations , i use the same conf file without any problems on squeeze (2.6.32-5) with .
ii conntrack 1:0.9.14-2 Program to modify the
conntrack tables

ii  conntrackd                         1:0.9.14-2 Connection tracking daemon
ii  libnetfilter-conntrack3            0.0.101-1

only a every two hours avg load peak activity.

Is this the same as the Debian bug #746464 [0]?

regards.

[0] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=746464

Signature · 1 line
Arturo Borrero González
#7 lore raw
In reply to Nicolas RENAULT
Quoted 12 lines · Nicolas RENAULT
General {
    Nice -20
    Scheduler {
        Type FIFO
        Priority 99
    }
    HashSize 32768
    HashLimit 131072
    LogFile on
    Syslog off
    LockFile /var/lock/conntrack.lock
    UNIX {
Quoted 8 lines
        Path /var/run/conntrackd.ctl
        Backlog 20
    }
    NetlinkBufferSize 2097152
    NetlinkBufferSizeMaxGrowth 8388608
    NetlinkOverrunResync On
    NetlinkEventsReliable Off
    PollSecs 15

I guess the assertion happens after 15 seconds.

conntrack-tools 1.2 includes expectation support which is broken with PollSecs. The code for polling expectations is missing. Thus, the assertion problem that you're reporting.

The problem is still there in 1.4.x, this needs a patch.