Re: [PLUG] Network boot link died...
On 03/09/2010 07:33 PM, Neal wrote: > Um, er, when we discussed this back in December it was pointed out > that the total cabling limit for 100BASE-CX and 1000BASE-CX (STP) is > 82 feet. Since you're also running an additional 100' of UTP in that > segment you're pushing your luck even without cable degradation and > the coupler in the run. > Um, according to the spec, the UTP/STP limit is 100 meters for 100bt and 1000bt. 100 meters is just over 328 feet Is it really STP? and is the shielding properly grounded? Russ ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
On Tue, Mar 09, 2010 at 11:06:28PM -0800, Michael Robinson wrote: > Should I jump up to Cat 6 STP cable? You should test both segments individually, and swap out the coupler in the middle. What that run is doing is NOT NORMAL. ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
Michael Robinson wrote: > If I understand the spec correctly for 100BaseTX, I should be fine. > The 90' cable is labeled as outdoor Cat 5e, though I wonder how > long is this cable going to last? > 100 Meters [300 ft] unshielded twisted-pair cable is the "standard" for *all* "baseband" runs. YMMV slightly depending on on cable choice [5, 5e, 6]. > I don't know what CX is and I'm not running a gigabit link, so the run > doesn't have to work at those speeds anyways. > CX = coaxial cable [cable tv] > Interestingly, turning off duplex and dropping the link speed to 10 mbps > stabilizes the link. I noticed that duplex was set to full on the link > and that the link was unstable with that setting. > f 10 Mbps half-dup works, but you can't get a stable connection that supports auto-negotiation to a higher speed and full-dup then there are 2 things to look at. 1. The cable run. 2. The NICs You can check for problems at both of these points by looking at ethernet error stats w. ethtool ethtool -S eth0 NIC statistics: rx_errors: 0 tx_errors: 0 tx_dropped: 0 collisions: 0 rx_length_errors: 0 rx_over_errors: 0 rx_crc_errors: 0 rx_frame_errors: 0 rx_missed_errors: 0 tx_aborted_errors: 0 tx_carrier_errors: 0 tx_fifo_errors: 0 tx_heartbeat_errors: 0 tx_window_errors: 0 tx_abort_late_coll: 0 tx_deferred_ok: 0 tx_single_coll_ok: 0 tx_multi_coll_ok: 0 tx_timeout_count: 0 tx_restart_queue: 0 rx_long_length_errors: 0 rx_short_length_errors: 0 rx_align_errors: 0 tx_tcp_seg_good: 0 tx_tcp_seg_failed: 0 rx_csum_offload_errors: 0 alloc_rx_buff_failed: 0 rx_dma_failed: 0 tx_dma_failed: 0 You should be able to just ifdown then ifup the interface to clear the counters and then watch the counters while both NICs are set to auto-negotiation. *The problem w. auto-neg is that not all vendors implement the standard correctly, so many times an Intel card and D-link card may not play well together. If you see a lot of collisions that could mean you have a bad NIC that's jabbering away and causing the other NIC t back-off and retransmit. If you see carrier errors that would be more indicative of a problem w. the cable. I don't know if this is a home-made or store bought cable. But the other thing you could try if you have the tools and skills and cut the ends of and re-crimp new connectors on them. *Especially if you see the protective sheath is not inside the connector, which means the hair-width copper wires can be easily bent and/or broken. ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
On Tue, 2010-03-09 at 19:33 -0800, Neal wrote: > Um, er, when we discussed this back in December it was pointed out > that the total cabling limit for 100BASE-CX and 1000BASE-CX (STP) is > 82 feet. Since you're also running an additional 100' of UTP in that > segment you're pushing your luck even without cable degradation and > the coupler in the run. > > Temporarily replace the STP cable with an above-ground substitute and > see if that clears things up. > > NealS > ___ > PLUG mailing list > PLUG@lists.pdxlinux.org > http://lists.pdxlinux.org/mailman/listinfo/plug If I understand the spec correctly for 100BaseTX, I should be fine. The 90' cable is labeled as outdoor Cat 5e, though I wonder how long is this cable going to last? I don't know what CX is and I'm not running a gigabit link, so the run doesn't have to work at those speeds anyways. Interestingly, turning off duplex and dropping the link speed to 10 mbps stabilizes the link. I noticed that duplex was set to full on the link and that the link was unstable with that setting. I suppose I can try a single run of UTP cable for the full 190'. That would eliminate the buried segment and the coupler as possible issues. The first 100' of Cat 5e UTP cable could be the problem. Trouble is, I have to bury the cable as that is the only practical way to go. Should I jump up to Cat 6 STP cable? Is there a special cable meant for direct burial that I should use? How about burying the line in conduit? I've heard that UTP cable can be used if in conduit burial is employed and moisture is kept out. Let me assume that the limit isn't 300 meters, let's say the limit is more like 82 feet. Well, why didn't putting a 10/100 switch where the coupler is fix the can't transmit faster than 10 mbps problem? Why was I able to buy a 100' Cat 5e UTP patch cable? ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
Um, er, when we discussed this back in December it was pointed out that the total cabling limit for 100BASE-CX and 1000BASE-CX (STP) is 82 feet. Since you're also running an additional 100' of UTP in that segment you're pushing your luck even without cable degradation and the coupler in the run. Temporarily replace the STP cable with an above-ground substitute and see if that clears things up. NealS ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
On Mon, Mar 08, 2010 at 08:40:01PM -0800, Michael Robinson wrote: > I have 100 feet of Cat 5e UTP cable wired from my server running into > an attic area. After that, a crossover coupler and 90 feet of Cat5e > STP cable follows. The client computer has a 100 mbps Netgear Fa311, > but the server is limited to 10 mbps because I historically could not > get a stable link at 100 mbps. OK, that's a bad sign right there. Your cable (or maybe NIC or switch) had a problem from the get-go, and now it's gone from 90% broken to 100% broken. It's probably your coupler, or some spot where the cable rubs against a sill or edge, or runs by something hot. Note also that you're only good for 300' in a single run of cable, even with no coupler in the middle. You're gonna have to test each run. Either one should be good for 1000 Mbps. ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug
Re: [PLUG] Network boot link died...
On Mon, Mar 8, 2010 at 8:40 PM, Michael Robinson wrote: > I have 100 feet of Cat 5e UTP cable wired from my server running into > an attic area. After that, a crossover coupler and 90 feet of Cat5e > STP cable follows. The client computer has a 100 mbps Netgear Fa311, > but the server is limited to 10 mbps because I historically could not > get a stable link at 100 mbps. Now I can't seem to get a stable link > at even 10 mbps. The 90' run goes outside, underground, and back > above ground into another building. I assume that the cable is burial > grade cable, but I'm starting to wonder. I keep getting ICMP time > exceeded packets. Is mixing shielded and unshielded cable the > problem? This worked stably yesterday and it has been working for > years. What can I tweak to deal with the timeouts or find out for > sure why the connection has gotten to be so weak? > I have had lots of problems with cat5 couplers. I have also had some that worked for years on end. But the ones I had trouble with was usually due to moisture. I suppose they would also be vulnerable to temperature. Is the coupler in a very insulated environment? I had one in my garage go bad. If at all possible, I would highly recommend replacing the entire run with a single cable. You can test this by just running a single long cable from end to end along the ground for a few minutes. It will either work, or it won't. -wes ___ PLUG mailing list PLUG@lists.pdxlinux.org http://lists.pdxlinux.org/mailman/listinfo/plug