Just 30 minutes back, I’d setup an rsync server. And thanks to Arindam Biswas who gave me the idea of using rsync for upstream transfer. Setting up an rsync server is very easy. Just create an
/etc/rsyncd.conf with following content:
uid = nobody gid = nobody use chroot = no [content] path = /path/to/content comment = Content
And, now start rsync in server mode with following command:
root@box# rsync --daemon
When done rsyncing kill the daemon process.
Well, this is cool. I accidentally clicked on a nntp link in Firefox, and then to my surprise instead of some newsreader application (like Evolution), Nautilus handled it. And then it downloaded recent
104 posts. When I double clicked on a posting, gedit has loaded that news posting. Isn’t it cool worth capturing ;-)
Yesterday was the day of mistakes. It all started with me submitting updates to my FreeBSD ports maintained by me: x11/libxcb and x11/xcb-proto. I initially submitted both PRs of both ports in
misc category. When I realized that I’ve posted in wrong category, then I reposted them into
ports category. But at the same time, some FreeBSD committer changed category of my old PRs from
misc resulting in duplicate PRs. Then they discarded the duplicate PRs. And then, within an 5-6 hours my port
x11/xcb-proto is committed and for
x11/libxcb, I received a bug report. Oops, I’d forgotten to include graphics/graphviz as
BUILD_DEPENDS, which is causing
doxygen to fail during documentation generation.
I just tried this great application named 0install, a cool way to deploy apps. No need to
su, just a
DnD, it will automatically download all dependencies and start the app. If I’m not mistaken, when I was living in properietary software world, there was a similar technology named ClickOnce introduced in .NET Framework 2.0 by Microsoft. Well there is similar technology for Java also, known as Java Web Start.
Anyways, I just installed zeroinstall-injector package from Fedora Extras repository. And then created a launcher for
0install on my
xfce4-panel. And now, from this 0installable software directory, I’ve chosen MiniClock to install, just DnDed its Zero install feed on the recently created
0launch launcher. Then it prompted me to trust GPG keys of the software maintainer (for security reasons), and after this it downloaded the dependency and I’m running MiniClock.
Now, to create a shortcut, I’ve to do:
0alias mini-clock http://rox4debian.berlios.de/0install/MiniClock.xml. This will create a shell (
/bin/sh) script named
~/bin directory, which will launch the application. Isn’t it cool :)
Well, this I just experienced a few minutes ago, and I think I should blog it. After queuing a mail on
127.0.0.1 smtp server in evening, I found that it was still not delivered. On
/var/log/maillog I found that
smtp port of
smtp.gmail.com is unreachable. I tried to telnet
smtps port, there I got connected. So, I thought Gmail might have dropped support for relaying mails via
smtp port, only
smtps port. So then I asked one of my friend Edwin, who is fortunately online from non-Indian (or to be more precise, non-Airtel) network, to connect to
smtp port of
smtp.gmail.com. He told me that he is able to connect. Now, the next thought that came in my mind, Gmail has blocked
smtp port of a specific box (though chances of that is very low, as these companies are running cluster of servers for load balancing, and anyways I never worked on clustered network :( ). So I tried another IPv4 address of
smtp.gmail.com, still same problem. Then I thought to try connecting to another SMTP server’s
smtp, so this time I tried MX of
microsoft.com, still same problem, so then I checked my box’s and my ADSL modem’s iptables, but no such rule found. So then I called Gautam to know if he is already experiencing this problem. So he also confirmed me the same. Then Edwin suggested me that I should try
tcptraceroute to see where packets were being dropped. Following is the stuff which I did:
[wahjava@chatteau ~]$ sudo tcptraceroute smtp.gmail.com -p 25 Password: traceroute to smtp.gmail.com (220.127.116.11), 30 hops max, 40 byte packets 1 172.16.0.1 (172.16.0.1) 1.503 ms 2.419 ms 3.376 ms 2 rasBTNLDel-static-18.104.22.168.mantraonline.com (22.214.171.124) 366.339 ms 367.289 ms 368.254 ms 3 rasBTNLDel-static-126.96.36.199.mantraonline.com (188.8.131.52) 378.217 ms 380.183 ms 384.148 ms 4 * * * 5 * * * 6 * * * 7 * * * 8 rasBTNLDel-static-184.108.40.206.mantraonline.com (220.127.116.11) 3340.447 ms !X * * [wahjava@chatteau ~]$ sudo tcptraceroute smtp.gmail.com -p 995 traceroute to smtp.gmail.com (18.104.22.168), 30 hops max, 40 byte packets 1 172.16.0.1 (172.16.0.1) 1.479 ms 2.408 ms 3.352 ms 2 rasBTNLDel-static-22.214.171.124.mantraonline.com (126.96.36.199) 20.312 ms 22.271 ms 23.240 ms 3 rasBTNLDel-static-188.8.131.52.mantraonline.com (184.108.40.206) 33.203 ms 34.168 ms 36.131 ms 4 rasBTNLDel-static-220.127.116.11.mantraonline.com (18.104.22.168) 48.099 ms 50.066 ms 51.032 ms 5 22.214.171.124 (126.96.36.199) 52.985 ms 53.949 ms 60.921 ms 6 188.8.131.52 (184.108.40.206) 84.881 ms 61.776 ms 65.554 ms 7 220.127.116.11 (18.104.22.168) 99.229 ms 101.843 ms 107.750 ms 8 22.214.171.124 (126.96.36.199) 103.664 ms 108.346 ms 3062.296 ms 9 188.8.131.52 (184.108.40.206) 3303.216 ms 303.740 ms 298.521 ms 10 220.127.116.11 (18.104.22.168) 321.219 ms 296.750 ms 297.529 ms 11 * * * 12 22.214.171.124 (126.96.36.199) 337.195 ms 295.753 ms 296.539 ms 13 188.8.131.52 (184.108.40.206) 362.315 ms 352.741 ms 354.526 ms 14 220.127.116.11 (18.104.22.168) 375.188 ms 22.214.171.124 (126.96.36.199) 378.140 ms 386.085 ms 15 188.8.131.52 (184.108.40.206) 379.977 ms 220.127.116.11 (18.104.22.168) 401.926 ms 402.796 ms 16 22.214.171.124 (126.96.36.199) 1969.214 ms 1969.046 ms 1969.025 ms [wahjava@chatteau ~]$ sudo tcptraceroute smtp.ncr.airtelbroadband.in -p 25 traceroute to smtp.ncr.airtelbroadband.in (188.8.131.52), 30 hops max, 40 byte packets 1 172.16.0.1 (172.16.0.1) 1.593 ms 2.507 ms 3.457 ms 2 rasBTNLDel-static-184.108.40.206.mantraonline.com (220.127.116.11) 398.424 ms 399.378 ms 403.343 ms 3 rasBTNLDel-static-18.104.22.168.mantraonline.com (22.214.171.124) 404.307 ms 406.274 ms 408.237 ms 4 rasBTNLDel-static-126.96.36.199.mantraonline.com (188.8.131.52) 414.203 ms 416.167 ms 435.135 ms 5 184.108.40.206 (220.127.116.11) 433.101 ms 434.066 ms 439.032 ms 6 * * * 7 smtp.airtelbroadband.in (18.104.22.168) 5374.574 ms 5374.184 ms 5374.113 ms [wahjava@chatteau ~]$ dig +short smtp.ncr.airtelbroadband.in 22.214.171.124 126.96.36.199 [wahjava@chatteau ~]$ telnet smtp.ncr.airtelbroadband.in smtp Trying 188.8.131.52... Connected to smtp.ncr.airtelbroadband.in (184.108.40.206). Escape character is '^]'. 220 **************************************************************** ^] telnet> quit Connection closed. [wahjava@chatteau ~]$ sudo tcptraceroute smtp.ncr.airtelbroadband.in -p 25 Password: traceroute to smtp.ncr.airtelbroadband.in (220.127.116.11), 30 hops max, 40 byte packets 1 172.16.0.1 (172.16.0.1) 1.434 ms 2.346 ms 3.297 ms 2 rasBTNLDel-static-18.104.22.168.mantraonline.com (22.214.171.124) 263.255 ms 268.210 ms 271.172 ms 3 rasBTNLDel-static-126.96.36.199.mantraonline.com (188.8.131.52) 270.139 ms 274.102 ms 276.069 ms 4 rasBTNLDel-static-184.108.40.206.mantraonline.com (220.127.116.11) 280.034 ms 280.991 ms 285.953 ms 5 18.104.22.168 (22.214.171.124) 287.920 ms 288.886 ms 290.852 ms 6 126.96.36.199 (188.8.131.52) 291.817 ms 321.753 ms 322.518 ms 7 smtpout.airtelbroadband.in (184.108.40.206) 973.533 ms 973.181 ms 973.111 ms
As clear from the above dump, that outgoing TCP packets to
smtp port are dropped at
220.127.116.11 allows packets to
smtp.ncr.airtelbroadband.in. So, what do you conclude from this ? Anyways, I reported this to their customer care, but customer care representative told me that they’re currently having some problem with some 3rd party websites. And he told me that issue will be resolved tomorrow. Let’s hope it get resolved soon.
Being a Fedora Core user, I’ve to use yum for installation, updation of software. yum is a cool package manager but it is quite irritating for me. Why ? Whenever you start yum, it implicitly checks for updates in repositories everytime you run it. That’s quite irritating if you’re on a low speed connection. I’ve used Ubuntu for some time, and used apt there. apt was my favorite package management tool, as you need to explicitly update repositories, but it is not supported very well on Fedora systems, as it need different style of repository (if I’m not mistaken).
And now few weeks back, when I was browsing Fedora Extras, I’ve seen this package manager, named smart . smart is a really smart package management system. It can connect to different kind of repositories. Currently, I’ve tried only
rpm-md repositories (aka yum repositories aka RPM meta-data repositories). So, no more yumming only smarting. Its configuration is quite easy. If you’re lazy (or dumb) when it comes to editing configuration files, you can also configure it on command line ;-) . It is able to download simultaneous 2-3 files. And there is not much difference in installing packages via smart, apt-get, or yum.To install package, do :
smart install package
To remove package, do:
smart remove package
To upgrade package, do:
smart upgrade package
To reinstall a package, do:
smart reinstall package
It is currently in beta testing, but I never experienced any problems with it. Happy smarting.
Today, I’ve visited Dad’s Surajmal Vihar office. They recently got an internet connection from Airtel with a static IPv4 address. So, I’ve to just check that internet connection for their Linux server. It was installed on a box with a proprietary OS. I figured out that it is installed in bridge mode, as the computer has a non-private IPv4 address manually set. The Airtel Broadband has provided them with a Zyxel Prestige 600 series modem.
I just plugged out the internet cable (actually a CAT5 cable with a RJ45 connector at both ends) from the box with proprietary OS and inserted in the
eth1 of the Linux server, and configured
eth1 with the static IPv4 address and appropriate gateway , but to my surprise I’m unable to ping the gateway :-(. Then I rechecked settings, and found them correct, and re-initialized
eth1. But still I’m unable to ping the gateway. I re-inserted cable into previous box ( with proprietary OS), and there I’m able to ping the gateway. Well I suspected that modem gets bound to a MAC address, as Gautam thrice told me that in his university, they can’t connect any other network equipment into their university’s network, as MAC addresses of all network devices have to be registered, as each port in the network switch is specifically bound to that MAC address. Anyways, I’m unable to figure out how to change settings in my modem, i.e. I don’t know what is my modem’s IPv4 address. So I just called the Airtel guy, who has installed the connection. He told me that modem is accessible at
192.168.1.1, and he is also surprised why modem is not responding when plugged into a Linux box. So, I replugged my internet cable into my Linux box. And modified
eth1 to be in
192.168.1.0/24 network. Then I pinged my modem. :-). And then I telnetted to the modem. And I found the modem interface quite complex as I’m a n00b in networking. So I simply exited from the modem interface without making any changes. And I restored
eth1 to the static IPv4 address and appropriate gateway. To my surprise it now worked :)
After some more disconnecting and reconnecting, I finally figured out that this modem gets binded to a MAC address and hence routes packet only for that source MAC address. And to change that binding, one has to just telnet (or probably just ping) the modem (i.e. 192.168.1.1 IPv4 address ) and modem will then record the new MAC address.
This is what I observed today. I don’t know if they did this for the security purpose, or it is just a bug. Chances are more that it is due to security purpose. Or is that some special bridging technique ? I’m confused :-S