TCP/IP quits about every other day over ADSL
Robert Basler
modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
days it quits talking to the internet, PING seems to indicate that
sometimes I can get packets out, but I can't get packets in, sometimes
PING indicates I can't even get packets out. I usually notice the
problem when I try to do a DNS from one of the other workstations and it
just sits there.
I have several workstations accessing the internet over the local LAN.
NAT is provided by Safefire Firewall, the workstation that I use the
most and the place where I usually notice the problem first is WSeB FP1.
I was previously running the original 16-bit TCP/IP stack and MPTS (I
upgraded both to 4.1 hoping to fix this problem but it remains.)
Rebooting the W4 box fixes the problem but this system is supposed to
run 24x7 and eventually I want to run a web server from it, so
unreliability is not an acceptable solution.
IP speed indicates that there is the standard keepalive traffic coming
from and going to the ADSL modem. The status lights on the modem are
also good.
Attempting to access the internet from the W4 box directly does not work
either.
At this point I'm suspecting the NAT or the firewall. Anyone else seen
anything like this?
Thanks.
Kris Kadela
Robert Basler wrote:
>
> I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
> modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
> days it quits talking to the internet, PING seems to indicate that
> sometimes I can get packets out, but I can't get packets in, sometimes
> PING indicates I can't even get packets out. I usually notice the
> problem when I try to do a DNS from one of the other workstations and it
> just sits there.
>
> I have several workstations accessing the internet over the local LAN.
> NAT is provided by Safefire Firewall, the workstation that I use the
> most and the place where I usually notice the problem first is WSeB FP1.
Same setup and same problems regardless if I run SafeFire or InJoy. I
have even changed the NICs a few times to no avail.
>
> I was previously running the original 16-bit TCP/IP stack and MPTS (I
> upgraded both to 4.1 hoping to fix this problem but it remains.)
>
> Rebooting the W4 box fixes the problem but this system is supposed to
> run 24x7 and eventually I want to run a web server from it, so
> unreliability is not an acceptable solution.
>
> IP speed indicates that there is the standard keepalive traffic coming
> from and going to the ADSL modem. The status lights on the modem are
> also good.
>
> Attempting to access the internet from the W4 box directly does not work
> either.
>
> At this point I'm suspecting the NAT or the firewall. Anyone else seen
> anything like this?
Like I said. Neither InJoy or SafeFire seem to have anything to do with
this. On another box while it doesn't quit I can't assign more than 3
IPs to it as the last one always gets bumped. I am beginning to look for
alternatives.
>
> Thanks.
--
Accretus Informatics
http://www.accretus.com
Edmonton and area personals
http://ep.accretus.com
Baden Kudrenecky
I have been having similar mini problems with one
computer screwing up the stack on my other one, but I
found that running 'setup' fixes everything. Until I
did that I was totally wierded out, as I could use
http, and receive POP3, but I couldn't SMTP or receive
NNTP.
baden
In <39AA9781...@bogusdomain.com>, Robert Basler <bo...@bogusdomain.com> writes:
>I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
>modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
>days it quits talking to the internet, PING seems to indicate that
>sometimes I can get packets out, but I can't get packets in, sometimes
>PING indicates I can't even get packets out. I usually notice the
>problem when I try to do a DNS from one of the other workstations and it
>just sits there.
>
>I have several workstations accessing the internet over the local LAN.
>NAT is provided by Safefire Firewall, the workstation that I use the
>most and the place where I usually notice the problem first is WSeB FP1.
>
>I was previously running the original 16-bit TCP/IP stack and MPTS (I
>upgraded both to 4.1 hoping to fix this problem but it remains.)
>
>Rebooting the W4 box fixes the problem but this system is supposed to
>run 24x7 and eventually I want to run a web server from it, so
>unreliability is not an acceptable solution.
>
>IP speed indicates that there is the standard keepalive traffic coming
>from and going to the ADSL modem. The status lights on the modem are
>also good.
>
>Attempting to access the internet from the W4 box directly does not work
>either.
>
>At this point I'm suspecting the NAT or the firewall. Anyone else seen
>anything like this?
>
>Thanks.
ba...@unixg.ubc.ca
http://baden.nu/
OS/2, Solaris & Linux
co...@catherders.com
>I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL modem
>hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of days it
>quits talking to the internet, PING seems to indicate that sometimes I
>can get packets out, but I can't get packets in, sometimes PING indicates
>I can't even get packets out. I usually notice the problem when I try to
>do a DNS from one of the other workstations and it just sits there.
>I have several workstations accessing the internet over the local LAN.
>NAT is provided by Safefire Firewall, the workstation that I use the most
>and the place where I usually notice the problem first is WSeB FP1.
>I was previously running the original 16-bit TCP/IP stack and MPTS (I
>upgraded both to 4.1 hoping to fix this problem but it remains.)
>Rebooting the W4 box fixes the problem but this system is supposed to run
>24x7 and eventually I want to run a web server from it, so unreliability
>is not an acceptable solution.
>IP speed indicates that there is the standard keepalive traffic coming
>from and going to the ADSL modem. The status lights on the modem are
>also good.
>Attempting to access the internet from the W4 box directly does not work
>either.
>At this point I'm suspecting the NAT or the firewall. Anyone else seen
>anything like this?
>Thanks.
Sounds like our configurations are almost identical (I use a different
aDSL modem and the injoy firewall/NAT.) Never had a problem like you
describe, though. Running weasel, apache, ftpd, telnetd, ntpd, and injoy
connect (ppp server) on a P2/200 w/256 and WSeB FP1.
I have to reboot every few weeks - I suspect a memory leak in either
apache or weasel, but I've been too busy to look into it. A cron job
reboots on sunday mornings.
Mike-
----------------------------------------------------------------
Please note: My public PGP key is available on my web site.
The email address is co...@catherders.com
The web page is at http://www.catherders.com
Because network administration is like herding cats.
----------------------------------------------------------------
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
Jack Troughton
>
> Robert Basler wrote:
> >
> > I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
> > modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
> > days it quits talking to the internet, PING seems to indicate that
> > sometimes I can get packets out, but I can't get packets in, sometimes
> > PING indicates I can't even get packets out. I usually notice the
> > problem when I try to do a DNS from one of the other workstations and it
> > just sits there.
> >
> > I have several workstations accessing the internet over the local LAN.
> > NAT is provided by Safefire Firewall, the workstation that I use the
> > most and the place where I usually notice the problem first is WSeB FP1.
>
> have even changed the NICs a few times to no avail.
I'm running over cable and using the packet filter in 4.1. I haven't
had this problem. Mind you, Netscape will toast me after a few days,
but when I'm not using it, I've had uptimes of over a month without
the problems you're describing.
When you have the problem, do all the machines on the lan stop
working, or only some of them? If all, look to the firewall and/or
provider, if not it's the specific boxes.
Of course, if I'm telling you something you know already, then you
may safely ignore me!:)
> > I was previously running the original 16-bit TCP/IP stack and MPTS (I
> > upgraded both to 4.1 hoping to fix this problem but it remains.)
> >
> > Rebooting the W4 box fixes the problem but this system is supposed to
> > run 24x7 and eventually I want to run a web server from it, so
> > unreliability is not an acceptable solution.
I can understand that!
> > IP speed indicates that there is the standard keepalive traffic coming
> > from and going to the ADSL modem. The status lights on the modem are
> > also good.
> >
> > Attempting to access the internet from the W4 box directly does not work
> > either.
> >
> > At this point I'm suspecting the NAT or the firewall. Anyone else seen
> > anything like this?
>
> Like I said. Neither InJoy or SafeFire seem to have anything to do with
> this. On another box while it doesn't quit I can't assign more than 3
> IPs to it as the last one always gets bumped. I am beginning to look for
> alternatives.
Hmmm... dunno about this one. My brother's running SafeFire and
doesn't see this either; his firewall is up in the two-month range
for planned uptime and it continues to work with no problems...
well, there's problems, but it's definitely screwed up routing on
the ISP's part; when we tracerte each other it always stops at
x.x.x.89 on my side, and x.x.x.90 on his... which is a reasonable
distance from our respective IPs.
I have an old ISA SMC card... I'm not completely sure what Tom's got
for a lan card, but it's an old ISA card as well... hold on, it's an
SMC 1208. Maybe you should try using tracerte instead of ping as a
diag; first to your gateway, then to your modem, then to your DNS
and see where it stops. If you can't reach the modem, try resetting
the modem.
Also, what fixlevel are you at? We're both running 8620; maybe it's
a case of a problem that's been fixed in a fixpack? Also, I've heard
that there are problems with 8621... maybe you need to downgrade.
Hmmm... also, when your system goes down, take a look at the routing
table on the machine that's down.
You know, I just thought of something; you might be running out of
(stack?) space in your routing table. In tcpip 4.1, there's a change
to how long routes live in the table, from 60 seconds to 7800
seconds. In my tcpexit.cmd (if it exists, it will be run
automatically after tcpstart.cmd) I have two lines I've added, to
change the keepalive time and to turn off multiple default routes.
Since 4.1 was envisioned in the first case as a server/router stack,
not a workstation stack, the defaults are set up to favour those
setups. Here are the lines:
inetcfg -set keepalive 60
inetcfg -set multidefrt 0
The first one set the time to live for routes to 60 seconds, and the
second tells the stack not to allow muliple default routes. Maybe
try that in tcpexit.cmd in \tcpip\bin, and see if it solves your
problem.
HTH!
Jack
Montreal PQ
CANADA
Robert Basler
Robert Basler
> I'm running over cable and using the packet filter in 4.1. I haven't
> had this problem. Mind you, Netscape will toast me after a few days,
Does TCP/IP 4.1 provide NAT? Or are you running a single workstation
connected to a cable modem? What do you mean by Netscape toasts you?
> When you have the problem, do all the machines on the lan stop
> working, or only some of them? If all, look to the firewall and/or
> provider, if not it's the specific boxes.
So far it looks like all, but I haven't turned additional boxes on after
things are stuck to see if they are stuck too.
> I have an old ISA SMC card... I'm not completely sure what Tom's got
> for a lan card, but it's an old ISA card as well... hold on, it's an
> SMC 1208. Maybe you should try using tracerte instead of ping as a
> diag; first to your gateway, then to your modem, then to your DNS
> and see where it stops. If you can't reach the modem, try resetting
> the modem.
That's a good suggestion, I'm going to figure out how to use tracerte
today.
> Also, what fixlevel are you at? We're both running 8620; maybe it's
> a case of a problem that's been fixed in a fixpack? Also, I've heard
> that there are problems with 8621... maybe you need to downgrade.
The WSeB box is TCP/IP 4.2, no fixpacks. The W4 box is TCP/IP 4.1 no
fixpacks.
> Hmmm... also, when your system goes down, take a look at the routing
> table on the machine that's down.
Second suggestion to do this.
> You know, I just thought of something; you might be running out of
> (stack?) space in your routing table. In tcpip 4.1, there's a change
> to how long routes live in the table, from 60 seconds to 7800
> seconds. In my tcpexit.cmd (if it exists, it will be run
> automatically after tcpstart.cmd) I have two lines I've added, to
> change the keepalive time and to turn off multiple default routes.
> Since 4.1 was envisioned in the first case as a server/router stack,
> not a workstation stack, the defaults are set up to favour those
> setups. Here are the lines:
>
> inetcfg -set keepalive 60
> inetcfg -set multidefrt 0
>
> The first one set the time to live for routes to 60 seconds, and the
> second tells the stack not to allow muliple default routes. Maybe
> try that in tcpexit.cmd in \tcpip\bin, and see if it solves your
> problem.
The netstat -r command gives me a ton of routes, despite the fact that
nothing had been done on the system since yesterday. Shouldn't it be
cleaning itself out if every 7200 seconds? I don't understand all this
yet, but below seems like a lot of entries:
destination router netmask metric flags
intrf
default 208.181.198.1 0.0.0.0 1
UGSP lan1
63.96.6.194 208.181.198.1 255.255.255.255 1
UGHW3 lan1
63.96.6.195 208.181.198.1 255.255.255.255 1
UGHW3 lan1
64.14.18.134 208.181.198.1 255.255.255.255 1
UGHW3 lan1
64.39.15.112 208.181.198.1 255.255.255.255 1
UGHW3 lan1
64.208.32.100 208.181.198.1 255.255.255.255 1
UGHW3 lan1
127.0.0.1 127.0.0.1 255.255.255.255 0
UH lo
128.11.45.101 208.181.198.1 255.255.255.255 1
UGHW3 lan1
128.11.45.110 208.181.198.1 255.255.255.255 1
UGHW3 lan1
128.11.45.151 208.181.198.1 255.255.255.255 1
UGHW3 lan1
128.123.3.50 208.181.198.1 255.255.255.255 1
UGHW3 lan1
142.146.40.100 208.181.198.1 255.255.255.255 1
UGHW3 lan1
168.144.91.167 208.181.198.1 255.255.255.255 1
UGHW3 lan1
168.144.124.211 208.181.198.1 255.255.255.255 1
UGHW3 lan1
192.44.75.30 208.181.198.1 255.255.255.255 1
UGHW3 lan1
192.168 192.168.0.100 255.255.255.0 0
UC lan0
192.215.73.34 208.181.198.1 255.255.255.255 1
UGHW3 lan1
192.215.73.37 208.181.198.1 255.255.255.255 1
UGHW3 lan1
192.215.73.71 208.181.198.1 255.255.255.255 1
UGHW3 lan1
193.99.144.71 208.181.198.1 255.255.255.255 1
UGHW3 lan1
194.158.96.140 208.181.198.1 255.255.255.255 1
UGHW3 lan1
194.175.173.4 208.181.198.1 255.255.255.255 1
UGHW3 lan1
195.41.1.69 208.181.198.1 255.255.255.255 1
UGHW3 lan1
195.41.1.118 208.181.198.1 255.255.255.255 1
UGHW3 lan1
195.54.80.70 208.181.198.1 255.255.255.255 1
UGHW3 lan1
195.154.220.34 208.181.198.1 255.255.255.255 1
UGHW3 lan1
199.60.228.51 208.181.198.1 255.255.255.255 1
UGHW3 lan1
199.60.229.9 208.181.198.1 255.255.255.255 1
UGHW3 lan1
199.60.229.16 208.181.198.1 255.255.255.255 1
UGHW3 lan1
199.93.4.187 208.181.198.1 255.255.255.255 1
UGHW3 lan1
204.174.19.36 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.65 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.72 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.74 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.77 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.81 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.181.112.84 208.181.198.1 255.255.255.255 1
UGHW3 lan1
205.189.31.195 208.181.198.1 255.255.255.255 1
UGHW3 lan1
207.126.101.92 208.181.198.1 255.255.255.255 1
UGHW3 lan1
207.136.80.39 208.181.198.1 255.255.255.255 1
UGHW3 lan1
207.194.200.1 208.181.198.1 255.255.255.255 1
UGHW3 lan1
207.194.200.18 208.181.198.1 255.255.255.255 1
UGHW3 lan1
207.194.200.129 208.181.198.1 255.255.255.255 1
UGHW3 lan1
208.181.198 208.181.198.1 255.255.255.0 1
UGSP lan1
208.181.198 208.181.198.240 255.255.254.0 0
UC lan1
208.184.216.203 208.181.198.1 255.255.255.255 1
UGHW3 lan1
208.184.216.224 208.181.198.1 255.255.255.255 1
UGHW3 lan1
209.68.34.246 208.181.198.1 255.255.255.255 1
UGHW3 lan1
209.75.22.238 208.181.198.1 255.255.255.255 1
UGHW3 lan1
209.85.28.135 208.181.198.1 255.255.255.255 1
UGHW3 lan1
209.85.157.6 208.181.198.1 255.255.255.255 1
UGHW3 lan1
213.160.64.51 208.181.198.1 255.255.255.255 1
UGHW3 lan1
216.32.16.74 208.181.198.1 255.255.255.255 1
UGHW3 lan1
216.32.16.75 208.181.198.1 255.255.255.255 1
UGHW3 lan1
216.167.78.68 208.181.198.1 255.255.255.255 1
UGHW3 lan1
Jack Troughton
>
> Jack Troughton wrote:
> > I'm running over cable and using the packet filter in 4.1. I haven't
> > had this problem. Mind you, Netscape will toast me after a few days,
>
> Does TCP/IP 4.1 provide NAT? Or are you running a single workstation
> connected to a cable modem? What do you mean by Netscape toasts you?
One workstation. Netscape 4.61 makes PM crash after being opened and
closed about a dozen times... try it for yourself and see! ;)
> > When you have the problem, do all the machines on the lan stop
> > working, or only some of them? If all, look to the firewall and/or
> > provider, if not it's the specific boxes.
>
> So far it looks like all, but I haven't turned additional boxes on after
> things are stuck to see if they are stuck too.
It might be a worthy experiment; it could help you narrow down the
possibilities of where the problem is...
> > I have an old ISA SMC card... I'm not completely sure what Tom's got
> > for a lan card, but it's an old ISA card as well... hold on, it's an
> > SMC 1208. Maybe you should try using tracerte instead of ping as a
> > diag; first to your gateway, then to your modem, then to your DNS
> > and see where it stops. If you can't reach the modem, try resetting
> > the modem.
>
> That's a good suggestion, I'm going to figure out how to use tracerte
> today.
tracerte hostname
It shows you each 'hop' to the destination, and 'pings' (not
exactly, but close enough to make no difference) each hop three
times.
> > Also, what fixlevel are you at? We're both running 8620; maybe it's
> > a case of a problem that's been fixed in a fixpack? Also, I've heard
> > that there are problems with 8621... maybe you need to downgrade.
>
> The WSeB box is TCP/IP 4.2, no fixpacks. The W4 box is TCP/IP 4.1 no
> fixpacks.
I don't know about 4.2, but definitely go to wr08620 on the 4.1. I
think that fix applies to 4.2 as well.
> > Hmmm... also, when your system goes down, take a look at the routing
> > table on the machine that's down.
>
> Second suggestion to do this.
Bad routing can completely bollix a network...
> > You know, I just thought of something; you might be running out of
> > (stack?) space in your routing table. In tcpip 4.1, there's a change
> > to how long routes live in the table, from 60 seconds to 7800
> > seconds. In my tcpexit.cmd (if it exists, it will be run
> > automatically after tcpstart.cmd) I have two lines I've added, to
> > change the keepalive time and to turn off multiple default routes.
> > Since 4.1 was envisioned in the first case as a server/router stack,
> > not a workstation stack, the defaults are set up to favour those
> > setups. Here are the lines:
> >
> > inetcfg -set keepalive 60
> > inetcfg -set multidefrt 0
> >
> > The first one set the time to live for routes to 60 seconds, and the
> > second tells the stack not to allow muliple default routes. Maybe
> > try that in tcpexit.cmd in \tcpip\bin, and see if it solves your
> > problem.
>
> The netstat -r command gives me a ton of routes, despite the fact that
> nothing had been done on the system since yesterday. Shouldn't it be
> cleaning itself out if every 7200 seconds? I don't understand all this
> yet, but below seems like a lot of entries:
Well, I've noticed that routes don't get discarded right away as you
think they would from the keepalive time...
<snip many routing entries;)>
This is much much longer than mine ever gets, and I have many cron
jobs running that require network activity, plus I'm on three or
four mail lists that have a _lot_ of traffic (all coming to my
address on the mail server I'm running on the box here). This could
be it... why don't you try making a tcpexit.cmd script with the two
lines I said above and see if it fixes your problems up?
Robert Basler
that if I run the SETUP.CMD script in MPTN\BIN as someone else suggested
that it comes back to life, so I'm going to have to figure out what part
of that script is fixing things then I should be able to figure out what
is going wrong. I notice that my computer was probed by crackers this
afternoon, so I'm going to keep an eye on that to see if there is any
coincedence to this.
Jack Troughton
wrote:
>I'll give that a try. I had it croak again this afternoon and found
>that if I run the SETUP.CMD script in MPTN\BIN as someone else suggested
>that it comes back to life, so I'm going to have to figure out what part
>of that script is fixing things then I should be able to figure out what
>is going wrong. I notice that my computer was probed by crackers this
>afternoon, so I'm going to keep an eye on that to see if there is any
>coincedence to this.
One of the things that setup.cmd does is flush the routing table...
the first line in mine is:
route -fh
which flushes the table.
Maybe try running that command by itself next time it croaks on you to
see if that works. Don't forget to set up the first default route
afterwards though!
>Jack Troughton wrote:
>> > > setups. Here are the lines:
>> > >
>> > > inetcfg -set keepalive 60
>> > > inetcfg -set multidefrt 0
>> > >
--
----------------------------------------------------------
* Jack Troughton jake at jakesplace.dhs.org *
* http://jakesplace.dhs.org ftp://jakesplace.dhs.org *
* Montréal PQ Canada news://jakesplace.dhs.org *
----------------------------------------------------------
Alex Taylor
>I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
>modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
>days it quits talking to the internet, PING seems to indicate that
>sometimes I can get packets out, but I can't get packets in, sometimes
>PING indicates I can't even get packets out. I usually notice the
>problem when I try to do a DNS from one of the other workstations and it
>just sits there.
Post the contents of your \MPTN\BIN\SETUP.CMD to the group. Several of us
have discovered that one line in that file can cause IP connectivity to
mysteriously die after a while with the 32-bit IP stacks. I think it's
the second 'route' statement that's superfluous, but post the file and
we'll take a look.
--
--------------------------
Alex Taylor
al...@eddie.cis.uoguelph.ca
--------------------------
Robert Basler
don't know what that does.
route -fh
arp -f
ifconfig lo 127.0.0.1
ifconfig lan0 192.168.0.100 netmask 255.255.255.0
ifconfig lan1 208.181.198.240 netmask 255.255.254.0
REM ifconfig lan2
REM ifconfig lan3
REM ifconfig lan4
REM ifconfig lan5
REM ifconfig lan6
REM ifconfig lan7
REM ifconfig sl0
route add default 208.181.198.1 -hopcount 1
route add -net 208.181.198 208.181.198.1 -netmask 255.255.255.0
-hopcount 1
ipgate on
Mark Dodel
Create a file called x:\TCPIP\BIN\TCPEXIT.CMD if you don't already
have one.
Add the following statements to the file:
INETCFG -S SYNATTACK 1
INETCFG -S keepalive 60
INETCFG -S multidefrt 0
The first sets up TCP/IP 4.1/4.2/4.3 to defend against ping floods.
The second reduces the amount of time a route remains in the ARP
table, thus reducing the route overflow which causes TCP/IP to freeze.
The last fixes the problem with TCP/IP 4.1/4.2/4.3 which can't handle
more then one ROUTE statement by default.
All of these have been listed as tips in past issues of the VOICE
Newsletter.
Mark
On Mon, 28 Aug 2000 16:46:57, Robert Basler <bo...@bogusdomain.com>
wrote:
-)I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
-)modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
-)days it quits talking to the internet, PING seems to indicate that
-)sometimes I can get packets out, but I can't get packets in, sometimes
-)PING indicates I can't even get packets out. I usually notice the
-)problem when I try to do a DNS from one of the other workstations and it
-)just sits there.
-)
-)I have several workstations accessing the internet over the local LAN.
-)NAT is provided by Safefire Firewall, the workstation that I use the
-)most and the place where I usually notice the problem first is WSeB FP1.
-)
-)I was previously running the original 16-bit TCP/IP stack and MPTS (I
-)upgraded both to 4.1 hoping to fix this problem but it remains.)
-)
-)Rebooting the W4 box fixes the problem but this system is supposed to
-)run 24x7 and eventually I want to run a web server from it, so
-)unreliability is not an acceptable solution.
-)
-)IP speed indicates that there is the standard keepalive traffic coming
-)from and going to the ADSL modem. The status lights on the modem are
-)also good.
-)
-)Attempting to access the internet from the W4 box directly does not work
-)either.
-)
-)At this point I'm suspecting the NAT or the firewall. Anyone else seen
-)anything like this?
-)
-)Thanks.
--
From the Desk of: Mark Dodel, RN, BSN, MBA
Healthcare Computer Consultant
mad...@ptdprolog.net http://home.ptd.net/~madodel
For a VOICE in the future of OS/2
http://www.os2voice.org/index.html
lsu...@mb.sympatico.ca
On Wed, 30 Aug 2000 14:07:25, Robert Basler <bo...@bogusdomain.com>
wrote:
> Here it is, the only thing I'm not sure I need is the IPGate part, I
> don't know what that does.
>
> route -fh
> arp -f
> ifconfig lo 127.0.0.1
> ifconfig lan0 192.168.0.100 netmask 255.255.255.0
> ifconfig lan1 208.181.198.240 netmask 255.255.254.0
> REM ifconfig lan2
> REM ifconfig lan3
> REM ifconfig lan4
> REM ifconfig lan5
> REM ifconfig lan6
> REM ifconfig lan7
> REM ifconfig sl0
> route add default 208.181.198.1 -hopcount 1
If you want to eliminate all the extra "routes" retained you
can change the line above to
route add default 208.181.198.1 netmask 255.255.255.255 -hopcount 1
*******
> route add -net 208.181.198 208.181.198.1 -netmask 255.255.255.0
The line above is not required and it causes problems
delete it
********
> -hopcount 1
> ipgate on
>
>
> Alex Taylor wrote:
> >
> > On Mon, 28 Aug 2000 12:46:57 -0400, Robert Basler <bo...@bogusdomain.com> wrote:
> > >I have ADSL with a fixed IP address (yes I'm sure) with a 3Com ADSL
> > >modem hooked to a Warp 4 FP 5 box running TCP/IP 4.1. Every couple of
> > >days it quits talking to the internet, PING seems to indicate that
> > >sometimes I can get packets out, but I can't get packets in, sometimes
> > >PING indicates I can't even get packets out. I usually notice the
> > >problem when I try to do a DNS from one of the other workstations and it
> > >just sits there.
> >
> > Post the contents of your \MPTN\BIN\SETUP.CMD to the group. Several of us
> > have discovered that one line in that file can cause IP connectivity to
> > mysteriously die after a while with the 32-bit IP stacks. I think it's
> > the second 'route' statement that's superfluous, but post the file and
> > we'll take a look.
> >
> > --
> > --------------------------
> > Alex Taylor
> > al...@eddie.cis.uoguelph.ca
> > --------------------------
--
Lorne Sunley
Kris Kadela
Mark Dodel wrote:
>
> This is a known problem with TCP/IP 4.1+
>
> Create a file called x:\TCPIP\BIN\TCPEXIT.CMD if you don't already
> have one.
>
> Add the following statements to the file:
>
> INETCFG -S SYNATTACK 1
> INETCFG -S keepalive 60
> INETCFG -S multidefrt 0
>
> The first sets up TCP/IP 4.1/4.2/4.3 to defend against ping floods.
>
> The second reduces the amount of time a route remains in the ARP
> table, thus reducing the route overflow which causes TCP/IP to freeze.
>
>
> The last fixes the problem with TCP/IP 4.1/4.2/4.3 which can't handle
> more then one ROUTE statement by default.
>
> All of these have been listed as tips in past issues of the VOICE
> Newsletter.
>
How would you go about setting this up (MPTS 8621, TCP/IP 4.1)?
Lan0, 4 IPs
I want 207.167.13.1 to be the default IP (BIND needs to use this IP)
and the other 3 as aliases.
The problem is that each new alias becomes the default IP and bind gets
screwed up.
Here is what happens exactly
ifconfig lan0 207.167.13.1 netmask 255.255.255.248 <-- becomes the
default according to iflist utility
after ifconfig 207.167.13.2 netmask 255.255.255.248 alias <-- becomes
the default
after ifconfig 207.167.13.3 netmask 255.255.255.248 alias <-- becomes
the default
after ifconfig 207.167.13.4 netmask 255.255.255.248 alias <-- becomes
the default
If I issue
ifconfig lan0 207.167.13.1 netmask 255.255.255.248 again the .3 becomes
the default and .2 gets dropped
if I issue ifconfig lan0 207.167.13.1 netmask 255.255.255.248 again then
.1 is the default and .2 is still gone
Now, I can't alias .2 again because it will become the default IP and
the whole setup is toast.
Either bind is screwed because it can't actually use the IPs specified
in the config (as of now it will only use the default IP no matter what
directive is in the config file) or TCP/IP's IP assignment is to blame.
Thomas Mueller
One workstation. Netscape 4.61 makes PM crash after being opened and
closed about a dozen times... try it for yourself and see! ;)
>
Why do you keep opening and closing Netscape, instead of leaving it open?
Netscape 4.04 is slow and clumsy with me, taking its sweet time opening and
closing, and I don't think I ever opened and closed Netscape more than two times
between reboots. Unfortunately, the whole system runs slower when Netscape is
open. Because of the time Netscape takes to open, I always open before going
online (56 K modem, so I am not always online). Once Netscape 4.04 bombed out
and took 15 minutes of hard-disk activity before it was done unloading.
So I use Netscape only when I need something special like https support.
I believe I would need a big Java 1.1.8 download if I choose to download
Netscape 4.61?
(Remove -nospam from e-dress)
Jack Troughton
>
> >
> One workstation. Netscape 4.61 makes PM crash after being opened and
> closed about a dozen times... try it for yourself and see! ;)
> >
>
> Why do you keep opening and closing Netscape, instead of leaving it open?
> Netscape 4.04 is slow and clumsy with me, taking its sweet time opening and
> closing, and I don't think I ever opened and closed Netscape more than two times
> between reboots. Unfortunately, the whole system runs slower when Netscape is
> open. Because of the time Netscape takes to open, I always open before going
> online (56 K modem, so I am not always online). Once Netscape 4.04 bombed out
> and took 15 minutes of hard-disk activity before it was done unloading.
I'm not the only one using my computer here. Family members use it
too, and it hasn't really sunk in when I ask them to not close
netscape.
> So I use Netscape only when I need something special like https support.
>
> I believe I would need a big Java 1.1.8 download if I choose to download
> Netscape 4.61?
No, java 1.1.8 is not a prereq for netscape. A 1.1 JVM is a prereq
if you want to use java in the browser, though.
Jack
Montreal PQ
CANADA
Alex Taylor
>Here it is, the only thing I'm not sure I need is the IPGate part, I
>don't know what that does.
The ipgate command enables IP forwarding. I'm not actually sure what it
does myself, but it's needed if your machine is a router or firewall...
not sure otherwise. Just leave it as is, is my advice.
As for the rest...
>route -fh
>arp -f
>ifconfig lo 127.0.0.1
>ifconfig lan0 192.168.0.100 netmask 255.255.255.0
>ifconfig lan1 208.181.198.240 netmask 255.255.254.0
>REM ifconfig lan2
>REM ifconfig lan3
>REM ifconfig lan4
>REM ifconfig lan5
>REM ifconfig lan6
>REM ifconfig lan7
>REM ifconfig sl0
>route add default 208.181.198.1 -hopcount 1
>route add -net 208.181.198 208.181.198.1 -netmask 255.255.255.0
^^^^^^^
'REM' out this line and see if the problem goes away.
Clive Dove
the Java runtime, you can forget the samples and the toolkit. If you
are in the US, you probably don't need the Unicode font.
So, download the Java 1.18 without the Unicode. Then get the August 31,
2000 fixes from Hursley Park.
On Sun, 3 Sep 2000 10:26:40, "Thomas Mueller"
<tmue...@bluegrass-nospam.net> wrote:
> > So I use Netscape only when I need something special like https support.
> >
> > I believe I would need a big Java 1.1.8 download if I choose to download
> > Netscape 4.61?
>
> >No, java 1.1.8 is not a prereq for netscape. A 1.1 JVM is a prereq
> if you want to use java in the browser, though.
>
> Jack
> Montreal PQ
> CANADA
> >
>
> I remember reading on the WWW (Warpcast, I believe) that Netscape 4.61 required
> Java 1.1.6, or was it 1.1.7? Latest Java I have is 1.1.4, installed from CD,
> which installed only the runtime and refused to install any of the other parts.
> So I see myself downloading Java 1.1.8, and then all but the runtime refuses to
> install. Sometime in the near future, Java 1.1.8 will become obsolete in favor
> of 1.3?
>
> Javascript seems much more prevalent than Java on the WWW, however both are a
> curse, so is Shockwave/Flash. Some web sites are prohibitively slow with any
> graphic browser (I use 56 K modem), and About and ZDNet are Lynx-only for me.
>
> (Remove -nospam from e-dress)
>
Robert Basler
of connectivity. Thanks!
Robert Basler
fills up and quits working to about 5 days. I have now added
inetcfg -s synattack 1
inetcfg -s keepalive 60
inetcfg -s multidefrt 0
to see if any of these help with the problem. The last one is the most
likely one since the other two don't deal with routes.
lsu...@mb.sympatico.ca
route add default 2.2.2.2 -netmask 255.255.255.255 -hopcount 1
Changing 2.2.2.2 to your own router/gateway address
I found the "netmask" keeps away the ever-expanding route
table blues.
Lorne
On Tue, 12 Sep 2000 20:10:50, Robert Basler <bo...@bogusdomain.com>
wrote:
--
Lorne Sunley
Robert Basler
all), however it stopped routing after about an hour of use.
Robert Basler
sockets" which I believe is the amount of time that a socket stays open
without traffic. It doesn't mention anything about routes.
With further fiddling my system now only runs for about an hour before
it stops routing. I think I'm going to go back a step or two and
regroup.
Jack Troughton wrote:
> > inetcfg -s synattack 1
> > inetcfg -s keepalive 60
> > inetcfg -s multidefrt 0
> >
> > to see if any of these help with the problem. The last one is the most
> > likely one since the other two don't deal with routes.
>
> Actually, the keepalive and multidedfrt both deal with routes;
> keepalive controls how long routes 'live' in the routing table...
Jack Troughton
>
> Taking out that route command has lengthened the time before the table
> fills up and quits working to about 5 days. I have now added
>
> inetcfg -s synattack 1
> inetcfg -s keepalive 60
> inetcfg -s multidefrt 0
>
> to see if any of these help with the problem. The last one is the most
> likely one since the other two don't deal with routes.
Actually, the keepalive and multidedfrt both deal with routes;
keepalive controls how long routes 'live' in the routing table...
> Robert Basler wrote:
> >
> > That looks to have been the key. I'm now on my third day without a loss
> > of connectivity. Thanks!
> >
> > Alex Taylor wrote:
> > > >route add -net 208.181.198 208.181.198.1 -netmask 255.255.255.0
> > > ^^^^^^^
> > > 'REM' out this line and see if the problem goes away.
Jack
Montreal PQ
CANADA
warpe...@my-deja.com
after some minutes to some subnets or internet through our LAN. (not a
dialup) My route table does not grow. PMtelnet sessions die. DNS is fine.
But will not connect. Some subnets answer pings, others don't.
I've not tried this.... But, would setting keepalive to 0, set it to infinity?
Regards,
Testing Deja's posting as I don't have the anti spam thing working :)
Hope this works...
In article <39BFA1F4...@bogusdomain.com>,
Robert Basler <bo...@bogusdomain.com> wrote:
> I tried that, it does indeed prevent the route table from growing (at
> all), however it stopped routing after about an hour of use.
>
> lsu...@mb.sympatico.ca wrote:
> >
Sent via Deja.com http://www.deja.com/
Before you buy.
Robert Basler
Robert Basler
192.168.55.100 before OS/2 stopped routing - I didn't think I had a
.55.100 device. OS/2 got stuck again today with only 25 routes in the
table. A little investigation has shown that this traffic comes from my
handheld Windows CE device when it is connected to the local NT4
machine. I have adjusted the firewall not to allow packets from it in
either direction now (it was configured just to prevent them from coming
in) and will report if that improves things. Coincidentally, the router
has worked the longest during the period when the Windows CE box was NOT
in use.
Does anyone else who is having this problem have a Wince device?
Turns out Windows CE uses addresses in the range 192.168.55.100-102 for
their own use.
warpe...@my-deja.com
see. I still have pmtelnet sessions dying after some minutes.
And cannot connect to some of our subnets without running the
setup.cmd.
That said. I would like some advice on returning this machine
to pre-tcp 4.1/4.3. Something like 4.02tqw or whatever.
Can someone give me some steps?
This has been very frustrating...
Thanks in advance
In article <39C10220...@bogusdomain.com>,
Alex Taylor
>That said. I would like some advice on returning this machine
>to pre-tcp 4.1/4.3. Something like 4.02tqw or whatever.
>Can someone give me some steps?
Only way to do that is to completely uninstall, remove, erase TCP/IP and
MPTS (and thus LAN services, Netware, and anything else that depends on
MPTS), and reinstall from the Warp GA CD. Then apply the 16-bit FixPaks.
AFAIK, anyway.
William L. Hartzell
Alex is right as everything else's updates depend upon the stack being
used.
Bill
warpe...@my-deja.com wrote:
>
> Well, I tried it, finally. No effect on my system, that I could
> see. I still have pmtelnet sessions dying after some minutes.
> And cannot connect to some of our subnets without running the
> setup.cmd.
>
> That said. I would like some advice on returning this machine
> to pre-tcp 4.1/4.3. Something like 4.02tqw or whatever.
> Can someone give me some steps?
>
warpe...@my-deja.com
the unistall/install a try:
I uninstalled tcp4.1 then installed tcp/ip and file and print client
from the GA CD. Inetver now is 4.00e. And, the dying sessions are
gone. I also reinstalled MPTS from the CD.
However, The requester service would not start as the ibmlan.ini
file had the familiar garbage in it for $tcpbeui. Fixed that. But,
The sub-service peer would not start(net3062).
I then ran peerinst.exe to fix that. Now, when I try to see
shares on the lan I get the sys0234 error "More data is available".
I had this fixed once before. But, don't remember exactly what I did.
I've been searching deja and I'm not sure this is a resolved issue.
I grabbed IP08402 and will give that a try. Net Use will
be the workaround until then. If anyone has more info on the
buffer issue, please post! Someone said that there was
and option in config.sys to increase the buffer from 4k. I can't
find where that is, though.
I sure hope the convenience pak gets all this straightened out!
As I am planning on a new system build in November!
What is depressing is I have since loaded tcp4.3 on another
system and it has been flawless! Luckily, I have no real need to
have the 32bit stack. I think jnapster led me down this path.
PMnapster is now my choice. :) I still plan to play with
SSH and IPsec in the future. I hope standards are in place
later this year so I can get a cable modem and use VPN.
Thanks for the input! I hope I can return the favor, someday.
Mike
In article <39CB2F94...@home.com>,