Resolved

Hi,

after upgrading from 1.22 to 1.23 I am experiencing connection aborts. For the first time it happened when I was online chatting with skype when skype suddenly went offline for some seconds. After some minutes the same break occured again. Later I discovered that the same event happened when browsing the Web (pages suddenly stopped loading). After a reload it worked again for about 10 minutes. I realized that setting "Idle timeout" from my default value of 10 minutes to 0 in the WAN configuration eliminates the problem (but that's no satisfying solution since I need to go offline automatically when the connection is not needed).

I also tried to download some Linux ISO images (using wget) from different servers. Again, when I set "Idle timeout" for example to 600 seconds, after 10 minutes the connection breaks and wget stops working until I stop/restart it. When deactivating idle timeout the download works as expected.

When downgrading m0n0wall (tested with 1.21 and 1.22 images) my router works without problems. The connection stays online during my download tests mentioned before. It then goes offline just when there is no traffic just as expected.

My WAN connection is build using PPPOE. I am using a WRAP board. I've executed up- and downgrades on m0n0wall's webinterface "Firmware" page. I didn't install 1.23 from scratch.

Bye,
fordprefect

i have de same problem i also use a wrap board and pppoe

Got the same problem with spontaneous disconnects on monowall 1.23.
As I have a flat rate internet connection, I do not use the "Dial on Demand" Feature.

OK folks, after an hour or so of trying to figure out what MPD does with all the hooks on the ng_bpf node that it creates, I think I've found the problem. It was introduced with the fix for the MSS fixup (oh the irony) and made MPD not count outgoing packets that match the on-demand criteria as it used to.

I've compiled an MPD version that hopefully fixes this on-demand problem; it's available at

https://neon1.net/temp/ondemandfix/mpd

Could you please install this as follows and see if it fixes your problem?

go to http://m0n0wall/exec.php
upload the fixed mpd file from the link above
execute the command "killall mpd"
execute the command "cp /tmp/mpd /usr/local/sbin"
click "Save" on the Interfaces: WAN page to restart MPD

The fix will be lost on reboot. If it works, I'll create an updated image (1.231 perhaps ;).

- Manuel

  works fine thank you for the fast fix

Hello, Manuel,

your fix works for me, too. Fine job.

Bye,
fordprefect

Thanks for the feedback - I'm glad to hear that it worked! The fix is now in the SVN repository; I'll wait for a week or two in case any other important fixes for 1.23 need to be incorporated, and then I'll release 1.231 with the fix.

- Manuel

Well done, Manuel.

Is this a problem if the Dial on Demand box is not ticked?

I think it is but just want to confirm.

Brian

Hello,

I may have a similar problem, so hopefully you guys can help me. I run a bunch of PHP scripts that connect to a lot of socket repeatedly using my ubuntu linux box. They connect to the same port on a particular server. This certain server is fine and is always reliable, and is used by many other people online, so I know they are probably not at fault. And the scripts worked great in 1.22.

But now that I've upgraded to 1.23 they seem to stall randomly when receiving the response from the server, making them take longer than before. The speed is the same as before, but I'll watch the scripts and it appears 1 of the scripts will always be hanging. It eventually gets the reply a few seconds later, but this stalling RARELY happened on 1.22.

I'm using the CD-ROM / Floppy combo of m0n0wall, I did a clean install (burned a new disc, used a newly formatted floppy). The machine is decent (5-6% cpu usage as the scripts run right now). None of the traffic is being blocked by the firewall. It's a Cable connection to the internet, and it has enough download/upload to handle the scripts too (as I said it worked great in 1.22). One funny thing I noticed is that the scripts used about 33 - 37KBps on 1.22, but now they only use about 24-30 KBps (which is fine cause it's using less bandwidth  , but this may be linked to the slowness of the scripts).

This isn't a huge deal, just bugging me since now they will take longer to finish  Any ideas or things to check? Thanks!

Hi, SomaFM,

does this behaviour also occur when you set "Idle timeout" in the WAN interface to 0? If not then there's a good chance that your scripts suffer from the bug described in this thread. Until Manuel releases a new version you can use his patch in the meantime.

Did you change your scripts since upgrading to 1.23? If so, I'd downgrade to 1.22 to check if the scripts still work flawlessly there.

Bye,
fordprefect

Hi, SomaFM,

does this behaviour also occur when you set "Idle timeout" in the WAN interface to 0? If not then there's a good chance that your scripts suffer from the bug described in this thread. Until Manuel releases a new version you can use his patch in the meantime.

Did you change your scripts since upgrading to 1.23? If so, I'd downgrade to 1.22 to check if the scripts still work flawlessly there.

Bye,
fordprefect

Thanks for the reply  My Cable connection uses DHCP so I have no option for Idle Timeout. I didn't change the scripts, but will go back to 1.22 and see how they run after the next batch is done (hate stopping these scripts cause it's a hassle to get them back where they left off  ). If 1.22 does the same thing, I will know it's a connection issue on my end, but if it works fine on 1.22 I will have to try out the quickfix mentioned above and see if that works. Either way, I'll reply with the results in a while. Thanks again!

m0n0wall 1.231 has just been released (http://m0n0.ch/wall/downloads.php) with this fix (although slightly altered to enable PPTP VPN idle timeout to work properly as well - see the change log).

Hey, great! Thank's, Manuel.

Bye,
fordprefect