Sean's blog

Bloggerunited.com down – will it return?

June 13th, 2009

Jess asked me what was going on with bloggerunited.com yesterday, as there’s currently only a domain-parking page there. I had a quick search online, and found LithiumMind’s blog that suggested it was suffering from bandwidth exhaustion last month. I often wondered how much bloggerunited was paying for bandwidth, as they must have used an awful lot!

The HTTP 509 (did you know 509 is not an official status code?) that LithiumMind saw has gone now, replaced by a domain parking page:

A domain parking page is bad news – who would use a webhost that displayed their own ads on your domain when you exceeded your bandwidth limits? That would be very cheeky! Perhaps whois can tell us a little more though. Whois is a tool that you can use to look up ownership, lease start / end, registrar info etc. For bloggerunited.com I see this:

Aiyah! Jess won’t be happy – bloggerunited.com was one of her favourite sites! See that ‘Expires On: 09-Jun-09’? We can only hope that there’s been a bit of an ‘admin oversight’ and the site will be back up soon.

Does anybody know the bloggerunited.com owner? Is it a brief holiday, or has the project run its course? I’d love to be able to tell Jess!

Update 17th June – seems like it could return, according to rei’s tweets it’s a credit card issue: http://twitter.com/rei99

Posted in Breaktime, Broken, Life | 1 Comment »

Streamyx DNS problems – Address Not Found

June 12th, 2009

I see other people posting complaints about TM’s DNS servers from time to time but had never suffered a problem specifically with DNS before this week. Since the weekend, browsing the Internet via Streamyx has been painful. Lots of websites simply don’t load.

I was a bit busy with something else, but my wife needed access to the Internet for her business, so at first I just switched her PC over to use our DiGi connection. Our DiGi connection is slower (because it uses the cellular network) than Streamyx, on one of Streamyx’s good days. On any of Streamyx’s very many bad days, the DiGi EDGE connection has never, ever let us down.

I had a look at the problem today and noticed that the browser (we use Firefox, but it would be the same in any browser) was failing to look up the domain name in the address. You can see the message about ‘Looking up…’ in the status line at the foot of the browser window:

What’s happening here is that the browser is looking up the Internet Address of ‘www.google.com.my’ from Streamyx’ DNS servers. The servers are not replying, and the browser eventually shows the ‘Address Not Found’ page to tell you the lookup has failed. Today this wasn’t a reliable problem, with lookups either failing, being very slow (several seconds), or succeeding as normal. Perhaps it’s a load problem, or perhaps it’s some other network issue. A ping (that I forgot to record) this afternoon showed some packet loss, but this evening as  I write this appears to be reliable.

While the problem was bad, I wondered if it might be a fault with the DNS servers. Many articles on the Internet recommend using OpenDNS‘ servers. In my mind, I’m already paying TM for DNS, so using a 3rd party service seems silly. I searched online to see if there were any other TM DNS servers I could use and found Tipvista.com’s List of Streamyx broadband backup DNS. At the time, I added the whole list and things seemed to be better – at least, there were no pages failing to load. I could see (using iptraf) that unanswered requests to the usual 2 Streamyx DNS servers (202.188.0.133 and 202.188.1.5) were being quickly followed by requests to other servers in the list, which were responding.

The list at tipvista isn’t all good for me. Some of those IP addresses do not respond to DNS queries, and some don’t do recursive DNS, so if they don’t already know the address for the domain you’re looking up, they won’t give a useful response. In fact, there’s only 1 in the list that does work for me: 202.188.0.132 – tested with:

dig www.example.com @202.188.0.132

You can use whois and dig to turn up a few more DNS servers from TM’s networks: whois will tell you what nameservers resolve queries for major TM websites, and dig allows you to test the nameservers you find to ensure they respond usefully to queries. For example:

whois streamyx.com

reports that two DNS servers, DNS1 and DNS2 in the jati.net.my domain resolve queries for streamyx.com – you can use the name of those servers directly, or ping them to see an IP address and use that.

Specifying fixed IP addresses for your DNS servers is a very brittle way of dealing with a problem that should either be solved by TM making their Streamyx DNS servers more reliable, or by offering DNS from a larger pool of servers. I wouldn’t really recommend this solution (of adding more DNS servers to your own config) to anyone, but then, what choice do we have in Malaysia?

Posted in Broadband, Broken | No Comments »

Pos Malaysia Parcel Tracking and Online feedback both broken

May 10th, 2009

A couple of customers of Lolyco.com have been asking this weekend when their orders will be delivered to Australia. When an order is shipped from Lolyco.com, an email is sent to the customer with a link to Pos Malaysia’s Parcel Tracking System and the Parcel Tracking Reference. First things first, Pos Malaysia have been 100% reliable for deliveries from Lolyco.com – we have no complaints there at all.

The customers who have contacted us recently have used the link to Pos Malaysia’s Tracking System, entered their Tracking Reference and got a message that says:

Your Reference Number does not exist!.

At which point, a customer might naturally conclude that we haven’t really posted their parcel. One of these orders was placed 10 days ago (6 working days + 1 public holiday). Time enough, one would imagine, to record a tracking reference.

We’re caught between a rock and a hard place with Pos Malaysia. Their nearest competitors charge around 5 times what Pos Malaysia charges for its cheapest delivery service. At one time we offered this competing service as a shipping option on the basis that it gave good quality tracking information. Not one single customer chose it.

We’re left with the final ironic option of telephoning Pos Malaysia to make an enquiry about their online parcel tracking system. We did try to use their feedback option, offered on the page that tells our customers that their tracking reference doesn’t exist, and got this reply by email:

Subject: Delivery Status Notification (Failure)

From: postmaster@pos.com.my

This is an automatically generated Delivery Status Notification.

Delivery to the following recipients failed.

       corpcomm@pos.com.my

So now you know. If the tracking system isn’t working, you can try the customer feedback system. That also doesn’t work!

Lolyco.com also maintains a couple of osCommerce shipping modules for integrating e-Commerce businesses with Pos Malaysia’s website. Pos Malaysia occasionally does something unnecessary and incompetent that breaks those too.

I went through a phase, a year or two ago, of emailing Pos Malaysia every few weeks offering to come to their office and fix their tracking system for free. I got no reply. I guess the upshot of this issue is this:

1. The tracking system is broken.
2. You can’t report that it’s broken because the reporting system is broken too.
3. They have no ability / desire to fix it.

I haven’t got anything more to say right now, unless someone knows how to spell the noise I would make if I went into the kitchen and boiled my head in a big pot.

Posted in Broken, Google | 14 Comments »

TM Net’s Streamyx. The productivity killer par excellence.

April 25th, 2009

I have never clicked on a link to slashdot.org and been impressed with the speed it loads at in Malaysia, it is usually more than five minutes before even a single page will display in my browser. Some days though, the pages never load at all. Today is one of those days. I just want to reply to a comment on slashdot, and all I get is the “Network Timeout – Try again” page from Firefox.

Here’s a ping to slashdot.org – 28% packet loss is completely normal for Streamyx, it is often this crap or worse:

sean@taiguima:~$ ping -n slashdot.org
PING slashdot.org (216.34.181.45) 56(84) bytes of data.
64 bytes from 216.34.181.45: icmp_seq=1 ttl=236 time=671 ms
64 bytes from 216.34.181.45: icmp_seq=2 ttl=236 time=670 ms
64 bytes from 216.34.181.45: icmp_seq=3 ttl=236 time=669 ms
64 bytes from 216.34.181.45: icmp_seq=4 ttl=236 time=660 ms
64 bytes from 216.34.181.45: icmp_seq=6 ttl=236 time=665 ms
64 bytes from 216.34.181.45: icmp_seq=7 ttl=236 time=658 ms
64 bytes from 216.34.181.45: icmp_seq=11 ttl=236 time=669 ms
64 bytes from 216.34.181.45: icmp_seq=13 ttl=236 time=672 ms
64 bytes from 216.34.181.45: icmp_seq=14 ttl=236 time=659 ms
64 bytes from 216.34.181.45: icmp_seq=15 ttl=236 time=672 ms
64 bytes from 216.34.181.45: icmp_seq=16 ttl=236 time=666 ms
64 bytes from 216.34.181.45: icmp_seq=17 ttl=236 time=672 ms
64 bytes from 216.34.181.45: icmp_seq=18 ttl=236 time=667 ms
64 bytes from 216.34.181.45: icmp_seq=19 ttl=236 time=655 ms
64 bytes from 216.34.181.45: icmp_seq=20 ttl=236 time=655 ms
^C
— slashdot.org ping statistics —
21 packets transmitted, 15 received, 28% packet loss, time 20040ms
rtt min/avg/max/mdev = 655.268/665.761/672.884/6.126 ms

I still have my DiGi EDGE adapter, but can’t paste a ping report: it seems as though DiGi have disabled ICMP on their network. DiGi never fails to work though, so even though it’s a slower network technology, I get the page I want from slashdot in under 30 seconds, reliably.

If you’re considering a web-browsing connection in Malaysia, I would strenuously recommend you avoid Streamyx like the plague. It’s not bad when it works, but in my experience those days are too few in number. Slow you can deal with. There’s absolutely nothing you can do with a broken Internet connection.

Posted in Broadband, Broken | No Comments »

A problem on DiGi EDGE at last!

April 15th, 2009

Just a quick note so I don’t forget all this stuff. I used DiGi OCS and D’Chat today to ask why I couldn’t access www.w3.org from my DiGi USB EDGE adapter. See the ping output below:

root@box0045:/home/sean# ping www.w3.org
PING www.w3.org (128.30.52.166) 56(84) bytes of data.
From crusher.w3.org (128.30.52.166) icmp_seq=1 Dest Unreachable, Bad Code: 10
From crusher.w3.org (128.30.52.166) icmp_seq=2 Dest Unreachable, Bad Code: 10
From crusher.w3.org (128.30.52.166) icmp_seq=3 Dest Unreachable, Bad Code: 10
From crusher.w3.org (128.30.52.166) icmp_seq=4 Dest Unreachable, Bad Code: 10
From crusher.w3.org (128.30.52.166) icmp_seq=5 Dest Unreachable, Bad Code: 10
^C
— www.w3.org ping statistics —
5 packets transmitted, 0 received, +5 errors, 100% packet loss, time 4021ms
, pipe 2

Pinging some sites is as reliable as ever, but I also get problems when I ping wikipedia:

root@box0045:/home/sean# ping en.wikipedia.org
PING rr.pmtpa.wikimedia.org (208.80.152.2) 56(84) bytes of data.
^C
— rr.pmtpa.wikimedia.org ping statistics —
9 packets transmitted, 0 received, 100% packet loss, time 8046ms

And when I ping www.tm.com.my:

root@box0045:/home/sean# ping www.tm.com.my
PING www.tm.com.my (202.71.108.103) 56(84) bytes of data.
64 bytes from 202.71.108.103: icmp_seq=1 ttl=115 time=884 ms
64 bytes from 202.71.108.103: icmp_seq=1 ttl=115 time=900 ms (DUP!)
64 bytes from 202.71.108.103: icmp_seq=2 ttl=115 time=662 ms
64 bytes from 202.71.108.103: icmp_seq=2 ttl=115 time=678 ms (DUP!)
64 bytes from 202.71.108.103: icmp_seq=3 ttl=115 time=214 ms
64 bytes from 202.71.108.103: icmp_seq=3 ttl=115 time=413 ms (DUP!)
64 bytes from 202.71.108.103: icmp_seq=4 ttl=115 time=370 ms
64 bytes from 202.71.108.103: icmp_seq=4 ttl=115 time=391 ms (DUP!)
^C
— www.tm.com.my ping statistics —
4 packets transmitted, 4 received, +4 duplicates, 0% packet loss, time 3010ms
rtt min/avg/max/mdev = 214.117/564.192/900.255/237.384 ms

A ping to bbcnews.com (for example) is as good as ever. Ping is not a reliable friend on DiGi’s network – I think they must have many network features clamped down for security and reliability, so I’m not convinced these commands would work reliably any other time. What I would expect though, is to be able to browse a web page at one of these sites. I get No Route To Host for w3.org:

root@box0045:/home/sean# curl -s -v www.w3.org
* About to connect() to www.w3.org port 80 (#0)
*   Trying 128.30.52.52… No route to host
*   Trying 128.30.52.51… No route to host
*   Trying 128.30.52.38… ^C

And SocketTimeout for wikipedia:

root@box0045:/home/sean# curl -s -v en.wikipedia.org
* About to connect() to en.wikipedia.org port 80 (#0)
*   Trying 208.80.152.2… Connection timed out
* couldn’t connect to host
* Closing connection #0

And for www.daemon.be:

root@box0045:/home/sean# curl -s -v www.daemon.be
* About to connect() to www.daemon.be port 80 (#0)
* Trying 78.47.186.210… Connection timed out
* couldn’t connect to host
* Closing connection #0

I forget the exact wording, but the nice person at DiGi tells me they’ll get a ‘data specialist’ (I think) to call me today or tomorrow. It’s 10pm now, so I guess tomorrow…

The funny thing is… when I called earlier, I could browse www.w3.org from my old Siemens MC60 using DiGi’s GPRS and a ‘digiwap’ APN, but it also failed from my wife’s posh smartphone using DiGi EDGE and ‘diginet’ APN. The DiGi D’Chat agent asked me what APN I was using for the USB adapter, and I couldn’t tell them! It’s not set anywhere in my pppd config files. I remember seeing how to discover it with an extended AT command, but it wasn’t until I got off the chat thing that I hunted it down:

at+cgdcont?
+CGDCONT: 1,"IP","diginet","0.0.0.0",0,0
+CGDCONT: 3,"IP","","0.0.0.0",0,0

OK

So my USB adapter is using the diginet APN, because I send the dial string “atd*99***1#” – I think!

I just tried plugging that USB EDGE adapter into my laptop and could fetch the robots.txt for en.wikipedia.org and www.dameon.be, but still the same problem with www.w3.org! With the USB EDGE adapter back in the server it is usually attached to, I still get the problem with www.w3.org, but can fetch the robots.txt from en.wikipedia.org and from www.daemon.be …?

Time to stop fiddling with this, for all I know there’s a guy in a DiGi uniform up my local radio mast, tongue sticking out of the corner of his mouth, plugging and unplugging a cable.

Update 28th April 2009 – Paka (see comment) reminds me this problem still exists. Here are today’s output from various utilites:

First ifconfig ppp0, just to prove we’re on DiGi’s network:

root@box0045:/home/sean# ifconfig ppp0
ppp0      Link encap:Point-to-Point Protocol
inet addr:10.32.64.240  P-t-P:192.200.1.21  Mask:255.255.255.255
UP POINTOPOINT RUNNING NOARP MULTICAST  MTU:1500  Metric:1
RX packets:123 errors:0 dropped:0 overruns:0 frame:0
TX packets:153 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:3
RX bytes:42772 (41.7 KiB)  TX bytes:17530 (17.1 KiB)

Lynx, a text-based browser (there’s no GUI on this machine):

sean@box0045:~$ lynx http://www.w3.org

Looking up www.w3.org
Making HTTP connection to www.w3.org
Alert!: Unable to connect to remote host.

lynx: Can’t access startfile http://www.w3.org/

ping (I suspect DiGi have made some changes to ICMP on their network recently, ping usually says “100% packet loss”):

sean@box0045:~$ ping www.w3.org
PING www.w3.org (128.30.52.53) 56(84) bytes of data.
From 210.48.192.158 icmp_seq=2 Packet filtered
From 210.48.192.158 icmp_seq=9 Packet filtered
From 210.48.192.158 icmp_seq=10 Packet filtered
From 210.48.192.158 icmp_seq=11 Packet filtered
From 210.48.192.158 icmp_seq=12 Packet filtered
From 210.48.192.158 icmp_seq=13 Packet filtered
^C
— www.w3.org ping statistics —
15 packets transmitted, 0 received, +6 errors, 100% packet loss, time 14023ms

and curl (“see URL”, a swiss-army knife of a utility, I use it usually to check headers from hosts):

sean@box0045:~$ curl -s -v http://www.w3.org/
* About to connect() to www.w3.org port 80 (#0)
*   Trying 128.30.52.54… No route to host
*   Trying 128.30.52.166… No route to host
*   Trying 128.30.52.168… No route to host
*   Trying 128.30.52.170… No route to host
*   Trying 128.30.52.38… No route to host
*   Trying 128.30.52.51… No route to host
*   Trying 128.30.52.52… No route to host
*   Trying 128.30.52.53… No route to host
* couldn’t connect to host
* Closing connection #0

But anything else seems to work just fine:

sean@box0045:~$ curl –head www.digi.com.my
HTTP/1.1 200 OK
Date: Tue, 28 Apr 2009 08:44:43 GMT
Server: Apache-Coyote/1.1
Content-Type: text/html;charset=ISO-8859-1
Set-Cookie: JSESSIONID=C58F88B5A8B39C5C5CF05C9C39B25EBF; Path=/
Connection: close

Time for D’Chat!

Posted in Broken | 3 Comments »