Huge packet loss ?

[petemoss]

To get to post this new topic, has taken me well over one hour (just to login and post).

Up to and including Monday night, my HSPA has been working flawlessly. High speeds (often 1.5 Mbps and sometimes up to 2 or even 3 Mbps). It would connect fine. Come Tuesday, I was able to connect, but the speed is slower than dialup, ... terrible. Of course web pages are out of the question, although tonight is the first time I have been able to even 'see' a web page. Here are the results of a ping to one of the Exetel dns IP's.

Code: Select all

$ ping 220.233.0.3
PING 220.233.0.3 (220.233.0.3) 56(84) bytes of data.
64 bytes from 220.233.0.3: icmp_seq=21 ttl=59 time=7617 ms
64 bytes from 220.233.0.3: icmp_seq=22 ttl=59 time=6611 ms
64 bytes from 220.233.0.3: icmp_seq=23 ttl=59 time=5603 ms
64 bytes from 220.233.0.3: icmp_seq=24 ttl=59 time=4601 ms
64 bytes from 220.233.0.3: icmp_seq=25 ttl=59 time=3595 ms
64 bytes from 220.233.0.3: icmp_seq=27 ttl=59 time=1727 ms
64 bytes from 220.233.0.3: icmp_seq=29 ttl=59 time=120 ms
64 bytes from 220.233.0.3: icmp_seq=30 ttl=59 time=99.0 ms
64 bytes from 220.233.0.3: icmp_seq=54 ttl=59 time=1610 ms
64 bytes from 220.233.0.3: icmp_seq=55 ttl=59 time=603 ms
64 bytes from 220.233.0.3: icmp_seq=56 ttl=59 time=93.4 ms
64 bytes from 220.233.0.3: icmp_seq=57 ttl=59 time=92.0 ms
64 bytes from 220.233.0.3: icmp_seq=58 ttl=59 time=5736 ms
64 bytes from 220.233.0.3: icmp_seq=120 ttl=59 time=9057 ms
64 bytes from 220.233.0.3: icmp_seq=121 ttl=59 time=8050 ms
64 bytes from 220.233.0.3: icmp_seq=122 ttl=59 time=7042 ms
64 bytes from 220.233.0.3: icmp_seq=123 ttl=59 time=6034 ms
64 bytes from 220.233.0.3: icmp_seq=128 ttl=59 time=1261 ms
64 bytes from 220.233.0.3: icmp_seq=129 ttl=59 time=256 ms
64 bytes from 220.233.0.3: icmp_seq=130 ttl=59 time=194 ms
64 bytes from 220.233.0.3: icmp_seq=131 ttl=59 time=105 ms
64 bytes from 220.233.0.3: icmp_seq=132 ttl=59 time=104 ms
64 bytes from 220.233.0.3: icmp_seq=133 ttl=59 time=103 ms
64 bytes from 220.233.0.3: icmp_seq=134 ttl=59 time=1631 ms
64 bytes from 220.233.0.3: icmp_seq=135 ttl=59 time=623 ms
64 bytes from 220.233.0.3: icmp_seq=136 ttl=59 time=121 ms
64 bytes from 220.233.0.3: icmp_seq=137 ttl=59 time=102 ms
^C
--- 220.233.0.3 ping statistics ---
146 packets transmitted, 27 received, 81% packet loss, time 145901ms
rtt min/avg/max/mdev = 92.005/2696.146/9057.157/3027.515 ms, pipe 9

I can send and receive emails, but only if they are very small, and POP3 has to be one account at one time, otherwise timeouts. I have not used email much, as an 8K msg takes nearly 10 mins to send. Downloaded some (Ubuntu) updates today, a sum of 1.1 Mb, and that took over 3 hours. There are often timeouts.

My setup is Ubuntu with an E169 modem (plugged direct, no router), and an external yagi antenna. This has worked fine for many months. Nothing has been changed on Ubuntu either. To prove that, I booted to XP Pro, and exactly the same problems showed up there, terrible speeds. My situation is not isolated either, as our neighbours (about 100 metres away) use Exetel HSPA, and they said they have the same problem. It happened within the same time frame, and they cannot even connect, I think.

We can probably rule out the weather. There have been no storms or such like for well over a week.

I have emailed support, but they have been of no help at all. I guess that's the problem with off shore support, the problem can't be dealt with first hand, or locally. Any remote support has to 'fly blind' of course, and is very limited.

Possibly this is a Optus tower problem ? Who can find out ?

It's 10:27 pm now, hopefully this post will be 'done' by midnight, ... here goes ....

Pete

[JasonM]

Wow, I thought you were exaggerating..
107 packets transmitted, 69 received, 35% packet loss, time 106043ms
rtt min/avg/max/mdev = 152.673/12063.126/32245.036/9627.563 ms, pipe 33

Guess not..

Do you know if the light on the modem is green, blue or cyan ?

[Affinity]

Hi Jason,

Uggh, sorry, editing this response.

Peter has seen the modem vary it's connection, it won't hold for higher speeds and keeps dropping back to GPRS/GSM. His neighbor has the same problem.

We suspect that the local tower has the issues as it did when a bushfire took it out previously; bit it has been working fine for months now as Peter said.

Kind Regards
AndrewM

[Affinity]

On the lights changing on the modem, as the speed is so slow, I'm able to see
what it's doing. I'm trying to load one web page (exetel post info), and there
was nothing being sent/received whilst the light was on blue, then I saw the
light go to cyan, and some data was sent/received, albeit not much.

The diff. lights are changing from one network to another.

I think blue is WCDMA, and cyan is HSPA ??

[petemoss]

Hi Jason,

For some reason I have been able to login today, and it didn't take ages, only about 5 mins.

As Affinity stated, the light is going from cyan on full for a considerable period, then the blue comes on for a few seconds, then back to cyan.

Seems it is swapping network types a lot.

The speed is still terrible, dialup would still be quicker no doubt. I _think_ there maybe a dns issue as well, not too sure.

Pete

[petemoss]

You know what it is like when you pay the power bill, all of a sudden, ..... whoosh, we have 'power' again. For some weird reason, all of a sudden, I'm getting good speeds ??????

Exele DSL speed test ....

Your current bandwidth reading is:

1.19 Mbps

which means you can download at 152.9 KB/sec. from our servers.

Now, I haven't changed anything on my end ??

Pete

[JasonM]

There has been no other changes, but it's possible the supplier did network maintenance.

Please monitor it, and advise if it reoccurs.

[petemoss]

I spoke too soon, speed is terrible again, basically much less than 1K/sec.

[JasonM]

OK, we'll ask Optus to look into it, please email residentialsupport@exetel.com.au with a link to this forum thread.

[petemoss]

Jason, can you PLEASE instruct the International help desk to do as you have requested, that is ...

OK, we'll ask Optus to look into it, please email residentialsupport@exetel.com.au with a link to this forum thread.

They are giving me the usual runaround, by asking that I change the hspa config options. These have already been tried many times, and speed is still slower than dialup.

Pete

[JasonM]

Actually, they've escalated it and they've found out the coverage in that area is 3G Voice only on the 900Mhz band (no 2100Mhz coverage).

[Affinity]

The country broadband advertising was promoting coverage up to 30km from the tower with a Yagi -- Peter has this setup and I don't think he is that far away from the tower....

There is also the fact that speeds have been good for at least some period of time and now all of a sudden they are bad; this does point to local issues, but Peter's neighbor has the same problems -- neither he or the neighbor have changed anything at all. So.... that only leads back to the tower itself or the links to/from that tower [discounting wireless signal coverage which has proven to be sufficient].

There was a prior incident relating to tower issues from bushfires and now the tower isn't performing again, but this time without any fires.

[petemoss]

Actually, they've escalated it and they've found out the coverage in that area is 3G Voice only on the 900Mhz band (no 2100Mhz coverage).

The Optus maps have always shown our address as on street' HSPA coverage.

The country broadband advertising was promoting coverage up to 30km from the tower with a Yagi -- Peter has this setup and I don't think he is that far away from the tower....

Both Google maps and Whereis kept giving errors, timeouts I guess, as they usually don't give errors. Anyway, with the tower/s being where they are, and our Yagi being pointed exactly at that direction for most of this year, I would put our distance at 6.8 km from the nearest tower.

I did try using http://web.acma.gov.au/pls/radcom/regis ... dispatcher to get the actual address, but searching for 'optus' gave many pages, and that would have taken me days to read through. Anyway, the tower is approx. 7 km away, and our connection/speed has been very good for many months now, and then it 'died" overnight (next door neighbours did also).

Surely Optus can supply Exetel with tower information ? Is it down, maintenance work, etc ? What caused this huge packet loss, which of course is still happening, usually well over 80% packet loss in most cases.

Wireless broadband can't just "die" overnight, without there being some sensible logical explanation. I'd appreciate an explanation.

Pete

PS Exetel should at least refund me one full week of HSPA, considering it has technically been "unusable" at average less than 1K/sec

[petemoss]

The speed has come good again. This has happened a few times, but only for 10 minutes or so, and then it "dies" for hours. This time it is still very good, for nearly 30 minutes.

Just downloaded 65 updates, 100 Mb in 15 mins, so the speed is very good.

I hope it lasts.

.. and here is a ping, to see how much different it is, from the first post ...

$ ping 220.233.0.3
PING 220.233.0.3 (220.233.0.3) 56(84) bytes of data.
64 bytes from 220.233.0.3: icmp_seq=1 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=2 ttl=59 time=141 ms
64 bytes from 220.233.0.3: icmp_seq=3 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=4 ttl=59 time=143 ms
64 bytes from 220.233.0.3: icmp_seq=5 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=6 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=7 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=8 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=9 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=10 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=11 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=12 ttl=59 time=132 ms
64 bytes from 220.233.0.3: icmp_seq=13 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=14 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=15 ttl=59 time=131 ms
64 bytes from 220.233.0.3: icmp_seq=16 ttl=59 time=233 ms
64 bytes from 220.233.0.3: icmp_seq=17 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=18 ttl=59 time=133 ms
64 bytes from 220.233.0.3: icmp_seq=19 ttl=59 time=133 ms
^C
--- 220.233.0.3 ping statistics ---
19 packets transmitted, 19 received, 0% packet loss, time 18002ms
rtt min/avg/max/mdev = 131.123/139.190/233.136/22.331 ms

No packet loss at all.

Pete

[petemoss]

.. and it died again at 4 pm ?