Koozali.org: home of the SME Server

Legacy Forums => General Discussion (Legacy) => Topic started by: yelnic on April 09, 2005, 04:48:50 AM

Title: Throughput WAY too slow
Post by: yelnic on April 09, 2005, 04:48:50 AM: I just installed a P4 on a customers' site using the SME as a server-gateway (firewall). This machine was to replace a P3 running Caldera.

I started out with a 3com 3C905CX and a RealTek card and the web browsing was horribly slow. Thinking something was the matter with the NIC configuration, I ran out to Radio Shack (only place here) and picked up two Linksys LNE100TXs. Started out great, but after a while, slowed to a crawl. If I pinged 166.82.1.3 from the root prompt on the server, one return would be 38ms and the next one would be 13343ms (yes, that's 13K!!).

Anybody have any ideas?

Thanks
Title: Throughput WAY too slow
Post by: axcairns on April 09, 2005, 07:01:47 AM: Just to eliminate possible causes, have you tried bypassing the server-gateway and see if you have similar problems?

Allan
Title: Throughput WAY too slow
Post by: jackl on April 09, 2005, 01:34:53 PM: Yelnic

If you are sure the SME server is ok, check patch cords, points, and cabling. I have seen bad cable terminations cause problems like this.

Regards
Jack
Title: Throughput WAY too slow
Post by: electroman00 on April 09, 2005, 04:13:45 PM: yelnic

ifconfig

Would be good to post here.
Title: Throughput WAY too slow
Post by: yelnic on April 11, 2005, 06:28:36 PM: Quote from: "axcairns"
Just to eliminate possible causes, have you tried bypassing the server-gateway and see if you have similar problems?

Allan

Yes, I put the old Caldera firewall back in service and it picked right up.

Thanks.
Title: Throughput WAY too slow
Post by: yelnic on April 11, 2005, 06:30:23 PM: Quote from: "jackl"
Yelnic

If you are sure the SME server is ok, check patch cords, points, and cabling. I have seen bad cable terminations cause problems like this.

Regards
Jack

Cabling works fine with the old firewall (and the new router), so I don't think that's the issue, but thanks. I have also seen these kinds of problems with bad wiring.

Thanks.
Title: Throughput WAY too slow
Post by: yelnic on April 11, 2005, 06:38:09 PM: Quote from: "electroman00"
yelnic

ifconfig

Would be good to post here.

I agree. Unfortunately, in the interest of getting the customer up and running we installed a Linksys router and are trying to configure the SME for server only. I did a complete re-install with the 3C905CX and now I can't get the machine to even ping anything other than itself.

I also have an internal modem installed and the configuration script is showing an "unknown" adapter as well as the 3com. Is this the modem, and is that messing me up? We would like to use the SmerpFax panel.

I finally left the site at 6:00 am Saturday morning (after arriving at 8:00 am on Friday), so I want to re-evaluate what I did in the wee hours of the morning and make sure I wasn't too fried for the obvious.

Thanks.
Title: Throughput WAY too slow
Post by: the_mad_prof on April 11, 2005, 08:52:13 PM: I'm sure you've eliminated the problem, but I had problems with a home installation once where SME would sometimes appear in the net neighbourhood, sometimes not. Turned out to be a subnet misconfiguration. Also Windoze boxes can have probs when servers are changed, they can time out looking for a specific server that is long gone.

Your mobo doesn't have an onboard NIC you haven't disabled does it?
Title: Throughput WAY too slow
Post by: yelnic on April 28, 2005, 07:44:42 AM: Quote from: "yelnic"
Quote from: "electroman00"
yelnic

ifconfig

Would be good to post here.

I finally left the site at 6:00 am Saturday morning (after arriving at 8:00 am on Friday), so I want to re-evaluate what I did in the wee hours of the morning and make sure I wasn't too fried for the obvious.

Thanks.

I now have the machine on my site and have re-configured it to work with my network. Strange stuff is happening.

ifconfig -a:

eth0 Link encap:Ethernet HWaddr 00:0A:5E:4F:61:0D
inet addr:192.168.63.211 Bcast:192.168.63.255 Mask:255.255.255.0
EtherTalk Phase 2 addr:65280/83
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:455 errors:3 dropped:0 overruns:244 frame:4
TX packets:323 errors:20 dropped:0 overruns:0 carrier:0
collisions:24
RX bytes:103045 (100.6 Kb) TX bytes:31227 (30.4 Kb)

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
EtherTalk Phase 2 addr:0/0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:32 errors:0 dropped:0 overruns:0 frame:0
TX packets:32 errors:0 dropped:0 overruns:0 carrier:0
collisions:0
RX bytes:3056 (2.9 Kb) TX bytes:3056 (2.9 Kb)

Here is the result of a ping to 166.82.1.3 (DNS):

PING 166.82.1.3 (166.82.1.3) from 192.168.63.211 : 56(84) bytes of data.
From 192.168.63.211 icmp_seq=10 Destination Host Unreachable
From 192.168.63.211 icmp_seq=11 Destination Host Unreachable
From 192.168.63.211 icmp_seq=12 Destination Host Unreachable
From 192.168.63.211 icmp_seq=13 Destination Host Unreachable
From 192.168.63.211 icmp_seq=14 Destination Host Unreachable
From 192.168.63.211 icmp_seq=15 Destination Host Unreachable
From 192.168.63.211 icmp_seq=16 Destination Host Unreachable
From 192.168.63.211 icmp_seq=17 Destination Host Unreachable
From 192.168.63.211 icmp_seq=18 Destination Host Unreachable
64 bytes from 166.82.1.3: icmp_seq=5 ttl=242 time=18560 ms
64 bytes from 166.82.1.3: icmp_seq=6 ttl=242 time=17559 ms
64 bytes from 166.82.1.3: icmp_seq=7 ttl=242 time=16560 ms
64 bytes from 166.82.1.3: icmp_seq=8 ttl=242 time=15560 ms
64 bytes from 166.82.1.3: icmp_seq=19 ttl=242 time=20459 ms
64 bytes from 166.82.1.3: icmp_seq=20 ttl=242 time=19450 ms
64 bytes from 166.82.1.3: icmp_seq=21 ttl=242 time=18450 ms
64 bytes from 166.82.1.3: icmp_seq=22 ttl=242 time=17450 ms
64 bytes from 166.82.1.3: icmp_seq=23 ttl=242 time=16450 ms
64 bytes from 166.82.1.3: icmp_seq=24 ttl=242 time=15450 ms
64 bytes from 166.82.1.3: icmp_seq=25 ttl=242 time=14450 ms
64 bytes from 166.82.1.3: icmp_seq=26 ttl=242 time=43450 ms
64 bytes from 166.82.1.3: icmp_seq=27 ttl=242 time=30210 ms
64 bytes from 166.82.1.3: icmp_seq=28 ttl=242 time=29200 ms
64 bytes from 166.82.1.3: icmp_seq=29 ttl=242 time=28200 ms
64 bytes from 166.82.1.3: icmp_seq=30 ttl=242 time=27200 ms
64 bytes from 166.82.1.3: icmp_seq=31 ttl=242 time=26200 ms
64 bytes from 166.82.1.3: icmp_seq=32 ttl=242 time=25200 ms
64 bytes from 166.82.1.3: icmp_seq=33 ttl=242 time=24200 ms
64 bytes from 166.82.1.3: icmp_seq=34 ttl=242 time=15760 ms
64 bytes from 166.82.1.3: icmp_seq=35 ttl=242 time=14750 ms
64 bytes from 166.82.1.3: icmp_seq=36 ttl=242 time=13750 ms
64 bytes from 166.82.1.3: icmp_seq=37 ttl=242 time=12750 ms
64 bytes from 166.82.1.3: icmp_seq=38 ttl=242 time=11750 ms
64 bytes from 166.82.1.3: icmp_seq=39 ttl=242 time=10750 ms
64 bytes from 166.82.1.3: icmp_seq=40 ttl=242 time=9750 ms

[ here it hung for about 10 minutes until I interrupted it with a
ctrl-C. ]

Any ideas? I can't get this machine to communicate consistently.

Thanks.
Title: Throughput WAY too slow
Post by: jackl on April 28, 2005, 09:46:22 AM: Yelnic,

eth0 is showing errors if it's not the patchcords/cabling have you tried a different nic.

Regards
Jack
Title: Throughput WAY too slow
Post by: yelnic on April 28, 2005, 08:22:56 PM: Quote from: "jackl"
Yelnic,

eth0 is showing errors if it's not the patchcords/cabling have you tried a different nic.

Regards
Jack

Yes. I am currently using a 3com 3C905CX-TX, but have used a Linksys LNE100TX with the same results.

Different NICs, different cables, different network --same results. <scratching my head>

Thanks.
Title: Throughput WAY too slow
Post by: Quail_Linux on April 28, 2005, 09:06:03 PM: Quote from: "yelnic"
Yes. I am currently using a 3com 3C905CX-TX, but have used a Linksys LNE100TX with the same results.

Different NICs, different cables, different network --same results. <scratching my head>

Thanks.

Hi,

(1) Have you by any chance tried the NIC in another PCI slot to see if that makes any differance ?
(2) Another thing to do is clean the contacts on the NIC's with a eraser.
(3) And check the PCI slot to make sure their is no hair or dust in the PCI slot.

Just couple of ideas

HTH
Title: Throughput WAY too slow
Post by: yelnic on April 28, 2005, 11:48:27 PM: Quote from: "Quail_Linux"

Hi,

(1) Have you by any chance tried the NIC in another PCI slot to see if that makes any differance ?
(2) Another thing to do is clean the contacts on the NIC's with a eraser.
(3) And check the PCI slot to make sure their is no hair or dust in the PCI slot.

Just couple of ideas

HTH

Argh. I moved the 3C905 to the top PCI slot and everything is working fine. Well, that's about 14 hours I won't get back.

Thanks for everyone's help.
Title: Throughput WAY too slow
Post by: smeghead on April 29, 2005, 06:38:44 AM: hmm, sounds like the card might be a bus mastering card and the slot you had it in wasn't enabled for busmastering; motherboard IRQ control issues ensued generating the probs observed.

May pay to check the docs on the mobo and keep this in mind in the future; of course it might just be a faulty board :-)