|
||||||||
Manuel Kasper said: > On 13.08.2004 11:33 +0200, Luc Naus wrote: > >> I have upgraded to 1.1b17 >> After having problems with my connection on version 1.0 and 1.1b16 >> (see mailinglist with item WRAP slows down), it seems to have been >> resolved in 1.1b16 with a hotfix from Fred Mol. >> I hoped it would be solved in 1.1b17, but unfortunatly it is not. > > Can you confirm that the "set pptp disable delayed-ack" option alone > makes the problem go away? The tcpmssfix really shouldn't make any > difference, as m0n0wall already does MSS clamping within ipnat. I've been testing several configurations: 1) 1.1b16 + disable delayed-ack + tcpmssfix Seems to work, but the general concensus is that tcpmssfix shouldn't make any difference 2) 1.1b16 + disable delayed-ack Seems to work. Has been running for two weeks without any problems. 3) 1.1b17 + disable delayed-ack Does NOT seem to work, unfortunately. Same problems showed up within 6 hours after reboot. 4) 1.1b17 + "set pptp disable windowing" This combination does seem to work. Has been running for 4 days without problems. It would be nice if other people having these problems could test configurations 3) and 4) to see whethere their findings are the same as mine. I stared at some tcpdumps after the last freeze and I noticed the same problems as before: packets from m0n0wall -> internet on the ng0 interface don't have corresponding packets on the interface that talks to the adsl modem. That's probably because of the following lines in ng_pptpgre.c: 477 /* Check if windowing is enabled */ 478 if (priv->conf.enableWindowing) { 479 /* Is our transmit window full? */ 480 if ((u_int32_t)PPTP_SEQ_DIFF(priv->xmitSeq, 481 priv->recvAck) >= a->xmitWin) { 482 priv->stats.xmitDrops++; 483 NG_FREE_M(m); 484 NG_FREE_ITEM(item); 485 return (ENOBUFS); 486 } 487 } Apparently when the packets are dropped, the transmit window is full, or at least the pptp module thinks it's full. By disabling windowing, packets are always sent, regardless of the transmit window. If we assume that the transmit window is not really full, but the whole flow control has somehow gotton out of sync, that option should fix the problem. I also noticed that the exact same change (disable delayed-ack -> disable windowing) was done in 1.1b17 for incoming (vpn) pptp connections. > If disabling delayed ACKs fixes it, I'll probably add an option to do > that to the WAN PPTP config section, because I'm not sure that > disabling it by default in all cases is a good idea (it might > seriously affect performance in other situations). > Maybe we need an option to disable windowing instead :-) > - Manuel > --Fred |