I'm getting this useless sources.list problem with netselect-apt too, I'm not sure what happened to my previous post. It's nearly identical to the posts above, it appears to work correctly, but it never spits out a single address, and it doesn't properly populate sources.list. I'm running wheezy, but I have to specify testing due to #623180. I did dig into the netselect-apt bash script, and I noticed a few funny things:
It seems nearly impossible to get the log() function to work within run_netselect() unless I force run_netselect() to fail, so troubleshooting is difficult. As far as I can tell, the first Perl inline command in net_select() creates a list of URLs to use correctly; i.e. $hosts is populated correctly. However, the call to the netselect command isn't working, and $out is null. Putting a "[ ! -z $out ] || exit -1" command just after the out= assignment always causes the entire netselect-apt script to exit and complain about a network error. I extracted the $hosts list, and ran netselect on it outside of netselect-apt. It seems every host on the list is reporting a TIMEOUT, so nothing gets selected. For whatever reason netselect-apt is ignoring this fact, and creating a useless (but not empty) sources.list. I tried downloading mirror_full from the main debian.org website and using it as the --infile option, but it suffers from the same problem. As far as I know, I'm not experiencing any general networking issues, both ping and traceroute to google com return as expected. Here's the raw netselect command I ran (hosts.test should be the contents of the $hosts shell variable): ~ # netselect -I -vvv -s 1 $(cat hosts.test | sed -e "s/\n/ /") Here's the tail of the output (stderr?) that netselect spits out: http://mirror.fdcservers.net/debian/ 9999 ms 30 hops 0% ok http://ftp.utexas.edu/debian/ 9999 ms 30 hops 0% ok http://debian.corenetworks.net/debian/ 9999 ms 30 hops 0% ok http://128.226.116.176/debian/ 9999 ms 30 hops 0% ok http://128.226.116.177/debian/ 9999 ms 30 hops 0% ok http://128.226.116.178/debian/ 9999 ms 30 hops 0% ok http://128.226.116.179/debian/ 9999 ms 30 hops 0% ok http://mirror.hmc.edu/debian/ 9999 ms 30 hops 0% ok http://mirrors.hosef.org/debian/ 9999 ms 30 hops 0% ok http://mirrors.modwest.com/debian/ 9999 ms 30 hops 0% ok http://mirrors.bloomu.edu/debian/ 9999 ms 30 hops 0% ok http://debian.usu.edu/debian/ 9999 ms 30 hops 0% ok http://debian.securedservers.com/debian/ 9999 ms 30 hops 0% ok http://debian.cc.lehigh.edu/debian/ 9999 ms 30 hops 0% ok http://debian.mirrors.easynews.com/linux/debian/ 9999 ms 30 hops 0% ok http://opensource.nchc.org.tw/debian/ 9999 ms 30 hops 0% ok http://mirrors.accretive-networks.net/debian/ 9999 ms 30 hops 0% ok http://debian.stream.uz/debian/ 9999 ms 30 hops 0% ok http://debian.unesr.edu.ve/debian/ 9999 ms 30 hops 0% ok http://mirror.rit.edu/debian/ 9999 ms 30 hops 0% ok http://mirror.its.uidaho.edu/pub/debian/ 9999 ms 30 hops 0% ok http://mirrors.tummy.com/debian/ 9999 ms 30 hops 0% ok http://ftp.th.debian.org/debian/ 9999 ms 30 hops 0% ok http://debian.mirror.ac.za/debian/ 9999 ms 30 hops 0% ok http://ftp.twaren.net/debian/ 9999 ms 30 hops 0% ok http://ftp.v6.coe.psu.ac.th/debian/ 9999 ms 30 hops 0% ok http://ftp.si.debian.org/debian/ 9999 ms 30 hops 0% ok http://149.20.4.71/debian/ 9999 ms 30 hops 0% ok http://149.20.20.135/debian/ 9999 ms 30 hops 0% ok Note there is no actual selection of a mirror. I did a quick scan of the entire output, and it appears that all of the hosts tested have the same results (i.e. 9999ms 30 hops 0% ok), even after netselect seems to cycle through the list at least twice. It's as if netselect is not doing either ping or traceroute correctly (as if it's trying a hybrid approach?), and totally failing silently. As an added bonus, the second inline Perl command seems to mangle this output ($out) just enough that netselect-apt doesn't notice the failure from netselect. This last bit is probably more suited to a netselect bug report, and #467303 does seem extremely related (I don't think it's appropriate to file this post on both bugs, but if you feel it necessary do so). Perhaps the root cause of the 9999ms is what we need to look at here. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
