Web lists-archives.com

Re: [Samba] Samba freezes

On 06/25/2017 11:34 PM, Andrew Bartlett via samba wrote:
> On Sun, 2017-06-25 at 17:54 +0100, Rowland Penny via samba wrote:
>> On Sun, 25 Jun 2017 18:11:05 +0200
>> Antoine Huchet via samba <samba@xxxxxxxxxxxxxxx> wrote:
>>> Hello,
>>> I have Samba 4.5.8-Debian running on Debian Stretch.
>>> I use Samba to load games over my local network from my Debian machine
>>> to my Playstation 2 (using OpenPS2Loader:
>>> https://github.com/Jay-Jay-OPL/OPL-Daily-Builds).
>>> From time to time, the game freezes, then after 5 minutes it comes
>>> back (sometimes it doesn't, depends on the game).
>>> And here is the syslog at the same time:
>>>> Jun 25 16:39:01 serveur CRON[23116]: (root) CMD (  [ -x
>>>> /usr/lib/php5/sessionclean ] && /usr/lib/php5/sessionclean)
>>>> Jun 25 16:41:05 serveur dhclient[717]: DHCPREQUEST of on
>>>> eth0 to port 67
>>>> Jun 25 16:41:05 serveur dhclient[717]: DHCPACK of from
>>>> Jun 25 16:41:05 serveur systemd[1]: Reloading Samba SMB Daemon.
>>>> Jun 25 16:41:05 serveur systemd[1]: Reloaded Samba SMB Daemon.
>>>> Jun 25 16:41:05 serveur dhclient[717]: bound to --
>>>> renewal in 2912 seconds.
>>> From what I see, there is the usual stuff on top of the logs, then it
>>> start echoing someting, then reloads and then it works again...
>>> Any idea what might be causing that?
>> It looks like your Samba machine is using DHCP and when it changes
>> systemd reloads Samba, this will break any connections.
>> Try giving your machine a fixed IP instead.
> G'Day Rowland,
> I think you are on to the right thing about the echo and DHCP, except
> they are in other order.  
> Antoine,
> What we do know is that the linux cifs client, and other clients no
> doubt, send an echo if they feel the server is un-responsive.  
> Is the DHCP renewal at the normal point (compare with previous
> renewal), or could this be both related to an underlying issue: lack of
> two-way link at the network layer?  Could Samba be listening only, but
> the client not getting the responses due to dodgy networking, and then
> it finally breaking down causing NetworkManager to re-try the whole
> stack?
> Can you rule out a dodgy network switch?
> Thanks,
> Andrew Bartlett

Thanks a lot for both your responses.

I tried assigning a static IP address to my server:
> allow-hotplug eth0
> #iface eth0 inet dhcp
> iface eth0 inet static
>         address
>         netmask
>         gateway
Now I don't get the DHCP message anymore.
So far, my PS2 hasn't frozen but I haven't tried for long enough to be
sure it is solved. I'll try again tomorrow.

I used to get the "Reloading Samba SMB Daemon" message every hour.
I only had one since 19:22:42 (4 hours and a half ago), which itself
shouldn't affect the PS2 because when I manually run `sudo service smbd
reload`, it doesn't affect the PS2.

> Jun 25 23:09:01 serveur CRON[2937]: (root) CMD (  [ -x
> /usr/lib/php5/sessionclean ] && /usr/lib/php5/sessionclean)
> Jun 25 23:12:22 serveur systemd[1]: Reloading Samba SMB Daemon.
> Jun 25 23:12:22 serveur systemd[1]: Reloaded Samba SMB Daemon.
> Jun 25 23:17:01 serveur CRON[2998]: (root) CMD (   cd / && run-parts
> --report /etc/cron.hourly)

It looks like the DHCP renewal was at the normal time. I haven't mesured
exactly but it seems correct to me. It was renewing approximately every
hour, that's about what the info messages said.

As for the lack of two way link at the network layer, I am not too sure
how to tell...

I only have one switch with everything on it. It is a TP-Link:
http://www.tp-link.gr/products/details/cat-42_TL-SG1024D.html . It works
well, I wouldn't call it dodgy but I could be wrong.



To unsubscribe from this list go to the following URL and read the
instructions:  https://lists.samba.org/mailman/options/samba