Half OT, question for Jeff

I have a DSL modem (Thomson aka SpeedtouchTG508) connected to my ISP, and between that and the other PCs in the house, a Netgear WGR614v9 wireless router. Recently it takes up to 5 minutes to logon (it used to take a minute, at most).

I copied and pasted the syslog from the DSL modem:

[Thu Aug 05 14:22:09 2010]:[SYS] Cold reboot [Thu Aug 05 14:22:09 2010]:[UPNP] UPnP daemon is ready to run [Thu Aug 05 14:22:10 2010]:[DHCPD] Recive REQUEST [Thu Aug 05 14:22:10 2010]:[DHCPD] no leases, packet->ciaddr:c0a80102 [Thu Aug 05 14:22:11 2010]:[DHCPD] sending ACK to 192.168.1.2 [Thu Aug 05 14:22:11 2010]:[DHCPD] unicasting packet to client ciaddr [Thu Aug 05 14:25:41 2010]:[ADSL] ATM bandwidth check failed. Adjustment is NEEDED !! [Thu Aug 05 14:25:42 2010]:[ADSL] max_pcr=2900 MUST
Reply to
Shadow
Loading thread data ...

4 min from start to finish. A bit long but not too horrible.

max_pcr (peak cell rate) has something to do with QoS at the ATM level. My guess(tm) is that the adaptive equalizer in the DSL modem is complaining that it can't get a decent signal and therefore insufficient bandwidth. Forget about the error log for now and check the Line level and the SNR (signal to noise ratio).

There might be a problem on the phone line or with whatever you're using for a DSL splitter or microfilter. Are you getting decent speeds after it connects?

Does it crap out when the phone rings?

Nope. This is at the lowest (ATM) level of the DSL modem, well below the firewall.

I don't know. Give me some signal quality numbers (line level and SNR).

Nope. Always blame someone else first. Then fix the problem.

Reply to
Jeff Liebermann

the error is here ...... nothing happens for over 3 minutes

I got a 11 minute lag this morning, before adjusting the pcr.

Link Information Uptime: 02:47:51 Modulation: ANSI T.1413 Line Status: Show Time DSL Type: Interleaved Path Bandwidth (Up/Down) [kbps/kbps]: 704 / 2304 Line Attenuation (Up/Down) [dB]: 6 / 12.5 SN Margin (Up/Down) [dB]: 21.0 / 40.0 CRC Errors (Up/Down): 0 / 13 HEC Errors (Up/Down): 0 / 3 I did try:

formatting link
With all phones unplugged: (other data was unchanged)

Line Attenuation (Up/Down) [dB]: 6 / 12.3 SN Margin (Up/Down) [dB]: 21.0 / 40.0 CRC Errors (Up/Down): 0 / 0 HEC Errors (Up/Down): 0 / 0

I finally managed to find a way to configure the pcr to

1660/1660/10, and I get no more errors. Still that 3-4 minute lag between the DHCP to the netgear and PPPoE routine starting up. (see below)
1.98Mb/s, which is what I pay for ... (2Mb/s)

Nope

...

Addendum: I managed to cut the time by a whole minute. I found that the Netgear was interpreting the ICMP packets from the modem as a DDoS attack(was in the logs), so I enabled "allow ping from internet" on the Netgear. I have them disallowed on the modem. Learnt something today. Thanks for the links. []'s PS but still don't understand why I had to lower the max_pcr value.

Reply to
Shadow

Looks very good. It's not signal levels.

Unplugging all the phones is one way to test for filter problems. However, it leaves the inside wiring in place. The wiring is often a problem. So, instead of unplugging everything except the DSL modem, I suggest you run a dedicated phone line to the NID (network interface device) and unplug the entire house at the NID. If it's too long a line, then just temporarily move the DSL modem near the NID and retest.

Weird. Could be two different problems. Note that AT&T supplied DSL modems in the USA do not allow that level of configuration. The settings pages are there in the modem firmware, but are not user accessible. Check with your ISP for a recommended set of operating parameters (which are not necessarily the default settings).

Weird. Thanks for the pointer. I'll look out for that.

Sorry, but I don't either. The value should be automagically set. Perhaps that DSL modem is guessing as to the correct value and takes few tries to make the determination?

Incidentally, a test of startup time on my Efficient 4100 DSL modem using fping:

(...)

About 32 seconds. Perhaps if you try a different modem...

Reply to
Jeff Liebermann

My time AFTER my netgear has established a connection with the modem:

[Tue Nov 15 16:58:30 2011]:[POE] ppp0 do disconnect I order modem to disconnect dsl then reconnect (via web interface) [Tue Nov 15 16:58:30 2011]:[PPP] ipcp down [Tue Nov 15 16:58:31 2011]:[POE] send PADT(4065) [Tue Nov 15 16:58:32 2011]:[UPNP] UPnP daemon is stopped [Tue Nov 15 16:58:33 2011]:[UPNP] UPnP daemon is ready to run [Tue Nov 15 16:58:33 2011]:[IGMP] forward multicast group 239.255.255.250. [Tue Nov 15 16:58:40 2011]:[POE] ppp0 do connect [Tue Nov 15 16:58:40 2011]:[POE] ppp0 connect mode=0 [Tue Nov 15 16:58:41 2011]:[POE] ppp0 start... [Tue Nov 15 16:58:41 2011]:[POE] ppp0 (VC0) send PADI [Tue Nov 15 16:58:42 2011]:[POE] ppp0 rcv PADO [Tue Nov 15 16:58:42 2011]:[POE] Send PADR, Enter PPPOE_SREQ [Tue Nov 15 16:58:43 2011]:[POE] ppp0 rcv PADS [Tue Nov 15 16:58:43 2011]:[POE] Enter PPPOE_CONNECTED, Session_ID=5372 [Tue Nov 15 16:58:44 2011]:[POE] ppp0 do connect [Tue Nov 15 16:58:44 2011]:[PPP] start lcp stage [Tue Nov 15 16:58:45 2011]:[PPP] ipcp up [Tue Nov 15 16:58:45 2011]:[PPP] Got gateway IP XXX.XXX.XXX.XXX [Tue Nov 15 16:58:46 2011]:[NTP] Sync with ptbtime1.ptb.de 16 seconds ..... Delay has something to do with the DHCP between netgear and modem. If a connection is already established, it's really fast. []'s

As to the max_pcr, I heard everyone in town (small town in interior of Brazil) had troubles connecting yesterday. Even the local WISP could not get their connection up. Apparently, I was the only one with a stable connection, after I altered the max_pcr. No idea if this was a coincidence. The ISP recommends 2900, I'm at 1660 Googling "max_pcr=2900" no quotes gives me just about every linux source code for dsl providers ........ Example:

formatting link
:) I also got this: PCR: Peak Cell Rate (PCR) is the maximum rate at which the sender can send cells. This parameter may be lower (but not higher) than the maximum line speed. 1 ATM cell is 53 bytes (424 bits), so a maximum speed of 832 Kbps gives a maximum PCR of 1962 cells/sec. This rate is not guaranteed because it is dependent on the line speed. SCR: Sustained Cell Rate (SCR) is the mean cell rate of a bursty, on-off traffic source that can be sent at the peak rate, and a parameter for burst-type traffic. SCR may not be greater than the PCR; the system default is 0 cells/sec. MBS: Maximum Burst Size (MBS) is the maximum number of cells that can be sent at the PCR. After MBS is reached, cell rates fall below SCR until cell rate averages to the SCR again. At this time, more cells (up to the MBS) can be sent at the PCR again. CBR is for connections that support constant rates of data transfer. The only parameter you need to worry about in CBR is PCR. UBR is for connections that have variable traffic. The only parameter you need to worry about in UBR is PCR. rtVBR is for connections that, while having variable traffic, require precise timing between traffic source and destination. PCR, SCR and MBS must all be set for rtVBR. nrtVBR is for connections that have variable traffic, do not require precise timing, but still require a set bandwidth availability. PCR, SCR and MBS must all be set for nrtVBR.

Reply to
Shadow

I have all my DSL modems setup for bridging with the PPPoE login/passwd in the router, not the modem. That eliminates the DHCP negotiation between the WAN side of the router router and modem.

Sigh. If it was everyone, then something probably changed in the DSLAM at the ATM level.

Hmmm... seems like someone decided to apply a rate cap to everyone. If you put max_pcr=2900 back, does it now work normally (no errors in the logs)?

Reply to
Jeff Liebermann

What I'm going to do.

[Wed Nov 16 16:11:42 2011]:[FW] **Port Scan Detected** 74.125.47.132

-> XXX.10.52.46, deny 74.125.47.132 for 5 mins

:) Evil

[Wed Nov 16 18:26:40 2011]:[ADSL] ATM bandwidth check failed. Adjustment is NEEDED !! [Wed Nov 16 18:26:40 2011]:[ADSL] max_pcr=2900 MUST
Reply to
Shadow

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.