https://bugzilla.novell.com/show_bug.cgi?id=824007 https://bugzilla.novell.com/show_bug.cgi?id=824007#c0 Summary: Serious IPV6 related trouble with Squid on 12.3/x86? (ipv6 turned off in yast) Classification: openSUSE Product: openSUSE 12.3 Version: Final Platform: i386 OS/Version: openSUSE 12.3 Status: NEW Severity: Normal Priority: P5 - None Component: Network AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: abittner@abittner.de QAContact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux i686; rv:21.0) Gecko/20100101 Firefox/21.0 Using a simple opensuse 12.3/x86 machine with squid-cache runnin, all clean install by opensuse media to make sure I am able to reproduce the problem. The whole network locally as well as the internet provider (consumer grade internet residential gateway (home router)) only uses ipv4 (pppoe on xdsl) I have in the recent months figured that with my production setup (12.3/x86 too) I often especially reproducable, couldnt surf to e.g. german news site http://www.stern.de, or actually the page would render in basic text-only like display, but all images missing. Firefox (running on the same opensuse 12.3/x86, proxy=localhost:8080) would display in its statusline that it was mostly waiting for d1.stern.de to reply (thats the server that delivers the images or some components of the page I guess). Another good test I came across is actually wikipedia.org, on the english pages, Firefox shows in its status line that its waiting for upload.wikimedia.org or so, where the images on Wikipedia articles come from. I think there were also some hints on the web that squid especially has trouble when also being used via SSL/TLS (https), thats especially true with wikipedia articles. So I reproduced this test setup, with clean 12.3/x86 and I am getting the same results. Latest patches to 12.3 applied as of today. Opensuse connects via default gateway in the LAN to the external world. http://www.stern.de stalling also happens for example when I set other local windows machines to use this quid on either the product suse 12.3 or the test 12.3. After reading much stuff in the past days, I finally came across hints about IPV6 trouble, although people not using ipv6 at all, and I ended up finding hints that its all Squids fault, and that users need to compile squid themselves without the ipv6 support in it, disabling it. I am trying to find out now, if this is really a squid issue, I have completely disabled the ipv6 in the general networking config in yast for both my test system and the production system of 12.3, both still show the problems. Then I additionally found some hints to run the local named/bind with the -4 parameter to make it only use ipv4, thats also set on both machines. Still doesnt help. So then I found the Squid related chatter, e.g. http://squid-web-proxy-cache.1019090.n4.nabble.com/squid-3-2-11-in-opensuse-... But I have found this kind of trouble with squid and ipv6 in general dating back as old as 2002 (not suse related, but giving hints that there is something seriously wrong with squid or ipv6 and all this compatibility mess). So is this really the state of the art, that todays standards are giving us users more and more trouble the more maybe DNS servers or Webservers and infrastructure move towards ipv6 and ipv4 mixed or badly configured or our ISPs are doing wrong things or whatever it is. Or is it the squid people that dont follow the rules? Or is this the best that engineering and people inventing and thinking about these things could come up with? I am really trying to find out if its me who is doing wrong things and if I can fix it somehow. There is no ipv6 stuff being displayed in ifconfig on eth0 or lo Sometimes these troubles calm down and stern.de gets usable for a while, but then again sometimes not or only half of the images or page loads, same with wikipedia/wikimedia. P.S. I used to have a bugreport about bind-related ipv6 problems in older opensuse versions, and I think there was a rather lenghty discussion involved or related later, with these layers further down into the operating system itself about glibc and what not, I dont remember that many details any more. Maybe thats an additional hint. Thanks. Reproducible: Sometimes Steps to Reproduce: 1. 2. 3. some varlogmessages excerpt: 2013-06-08T10:21:30.867686-04:00 linux-z0j0 squid[1855]: Process ID 1855 2013-06-08T10:21:30.867963-04:00 linux-z0j0 squid[1855]: Process Roles: worker 2013-06-08T10:21:30.868275-04:00 linux-z0j0 squid[1855]: With 4096 file descriptors available 2013-06-08T10:21:30.868572-04:00 linux-z0j0 squid[1855]: Initializing IP Cache... 2013-06-08T10:21:30.868886-04:00 linux-z0j0 squid[1855]: DNS Socket created at [::], FD 8 2013-06-08T10:21:30.869126-04:00 linux-z0j0 squid[1855]: DNS Socket created at 0.0.0.0, FD 9 some varlogwarn excerpts: 2013-06-08T10:21:30.867303-04:00 linux-z0j0 squid[1855]: Starting Squid Cache version 3.2.11 for i686-pc-linux-gnu... 2013-06-08T10:21:31.075648-04:00 linux-z0j0 squid[1855]: commBind: Cannot bind socket FD 16 to [::1]: (99) Cannot assign requested address 2013-06-08T10:21:31.076792-04:00 linux-z0j0 squid[1855]: commBind: Cannot bind socket FD 17 to [::1]: (99) Cannot assign requested address 2013-06-08T10:21:31.078708-04:00 linux-z0j0 squid[1855]: ERROR: Failed to create helper child read FD: UDP[::1] some varlogsquidcachelog excerpts 2013/06/08 10:21:30 kid1| Initializing IP Cache... 2013/06/08 10:21:30 kid1| DNS Socket created at [::], FD 8 2013/06/08 10:21:30 kid1| DNS Socket created at 0.0.0.0, FD 9 2013/06/08 10:21:31 kid1| HTCP Disabled. 2013/06/08 10:21:31 kid1| commBind: Cannot bind socket FD 16 to [::1]: (99) Cannot assign requested address 2013/06/08 10:21:31 kid1| commBind: Cannot bind socket FD 17 to [::1]: (99) Cannot assign requested address 2013/06/08 10:21:31 kid1| ERROR: Failed to create helper child read FD: UDP[::1] -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.