March 2013 - calendarserver-dev - lists.macosforge.org

Debian updated package 4.2
by Fredrik Unger 05 Mar '14

05 Mar '14

Hi, I updated my personal package to 4.2. https://tree.se/debian/ There were three minor changes, fixing a few SQL statements for 0.1.4, adding python-psutil as dependency, debianizing PGSOCKETDIR. I will improve on packaging, this was the first attempt. The server starts and http://localhost:8008/ works in the browser. Not tested any further. Not a SAFE upgrade path, and it might eat your cat. It is for testing on a empty server without any important data. With regards to the SQL patch. Looking at trunk some of these SQL statements are OK for 0.1.4, some are not. It is the same problems with 'primary key(' and 'unique(' as in Ticket #759. http://trac.calendarserver.org/ticket/759 Example from current trunk (r10297) is line 186 vs 192. http://trac.calendarserver.org/browser/CalendarServer/trunk/txdav/common/da… Still not a big deal and calendar server does not support 0.1.4, but 0.1.2 http://trac.calendarserver.org/browser/CalendarServer/trunk/support/build.s… Not sure it makes sense to open a bug against the SQL issue as new code got added with 'primary key(' sql. It is after all a sqlparse bug. Better for packagers to patch it I think, as the distribution is using the "wrong" version. To fix it is just to add a space, and the issues is now well known. Thanks /Fred

3 6

Re: [CalendarServer-dev] IPv6 support in 4.2?
by Glyph 26 Mar '13

26 Mar '13

Back to the list again... (Please remember to 'reply all'...) That config looks OK to me. Can you paste the exact log message, along with some surrounding context in the log? Regardless of what gets logged, is it listening on the IPv6 interface? Maybe this is another FreeBSD-specific issue, I don't think we've tested IPv6 there... -glyph On Mar 25, 2013, at 3:10 AM, Axel Rau <Axel.Rau(a)Chaos1.DE> wrote: > Am 24.03.2013 um 21:56 schrieb glyph(a)twistedmatrix.com: > >> Sorry, accidentally replied directly; back on-list... >> >> On Mar 24, 2013, at 1:21 AM, Axel Rau <axel.rau(a)chaos1.de> wrote: >> >>> >>> Am 24.03.2013 um 04:00 schrieb glyph(a)twistedmatrix.com: >>> >>>> >>>> On Mar 21, 2013, at 2:58 PM, Axel Rau <axel.rau(a)chaos1.de> wrote: >>>> >>>>> I see some IPv6 patches in backport, while getting >>>>> >>>>> twisted.internet.error.CannotListenError: Couldn't listen on dead:beef:f:d::cc:8443: [Errno 43] Protocol not supported. >>>>> >>>>> Is this expected to work in 4.2? >>>> >>>> Yes, but only if your host actually has an IPv6 address. Do you? >>> Yes, its cp4.lrau.net. >>> So, I will have to debug that on my FreeBSD port. I first check, if the patch code in backport is being called... >>> Any suggestions welcome... >> >> The backport is identical to the code in newer versions of Twisted, so it shouldn't matter. It would still be good to know for diagnostic purposes, of course. >> >>>>> If yes, does the TLS/cert code handle dual stack addresses? >>>> >>>> Sure, that's at a totally different layer. TLS certificates are for a hostname, not an IP. >>> My experience is, that some server/client implementations check reverse mapping and fail if there is more than one address per fqdn: >>> [axels-macpro:~] axel% host cp4.lrau.net >>> cp4.lrau.net has address 91.216.35.77 >>> cp4.lrau.net has IPv6 address 2a02:d40:2:2::77 >>> [axels-macpro:~] axel% host 91.216.35.77 >>> 77.35.216.91.in-addr.arpa domain name pointer cp4.lrau.net. >>> [axels-macpro:~] axel% host 2a02:d40:2:2::77 >>> 7.7.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0.2.0.0.0.0.4.d.0.2.0.a.2.ip6.arpa domain name pointercp4.lrau.net. >>> [axels-macpro:~] axel% ping6 cp4 >>> PING6(56=40+8+8 bytes) 2a02:d40:2:10::121 --> 2a02:d40:2:2::77 >>> 16 bytes from 2a02:d40:2:2::77, icmp_seq=0 hlim=62 time=18.691 ms >> >> I don't believe we have any such check. >> >> What relevant configuration keys do you have set? > > --- >  > <key>ServerHostName</key> > <string>cp4.lrau.net</string>  > >  > <key>HTTPPort</key> > <integer>0</integer> > >  >  > > <key>SSLPort</key> > <integer>8443</integer> > >  > <key>EnableSSL</key> > <true/> > > <key>SSLMethod</key> > <string>TLSv1_METHOD</string> > > <key>SSLCiphers</key> > <string>HIGH:MEDIUM</string> > >  > <key>RedirectHTTPToHTTPS</key> > <false/> > > >  > >  > <key>BindAddresses</key> > <array> > <string>91.216.35.77</string> > <string>2a02:d40:2:2::77</string> > </array> > >  > <key>BindHTTPPorts</key> > <array> > </array> > >  > <key>BindSSLPorts</key> > <array> > <integer>8443</integer> > </array> > --- > > Axel > --- > PGP-Key:29E99DD6 ☀ +49 151 2300 9283 ☀ computing @ chaos claudius >

2 2

Re: [CalendarServer-dev] IPv6 support in 4.2?
by Glyph 24 Mar '13

24 Mar '13

Sorry, accidentally replied directly; back on-list... On Mar 24, 2013, at 1:21 AM, Axel Rau <axel.rau(a)chaos1.de> wrote: > > Am 24.03.2013 um 04:00 schrieb glyph(a)twistedmatrix.com: > >> >> On Mar 21, 2013, at 2:58 PM, Axel Rau <axel.rau(a)chaos1.de> wrote: >> >>> I see some IPv6 patches in backport, while getting >>> >>> twisted.internet.error.CannotListenError: Couldn't listen on dead:beef:f:d::cc:8443: [Errno 43] Protocol not supported. >>> >>> Is this expected to work in 4.2? >> >> Yes, but only if your host actually has an IPv6 address. Do you? > Yes, its cp4.lrau.net. > So, I will have to debug that on my FreeBSD port. I first check, if the patch code in backport is being called... > Any suggestions welcome... The backport is identical to the code in newer versions of Twisted, so it shouldn't matter. It would still be good to know for diagnostic purposes, of course. >>> If yes, does the TLS/cert code handle dual stack addresses? >> >> Sure, that's at a totally different layer. TLS certificates are for a hostname, not an IP. > My experience is, that some server/client implementations check reverse mapping and fail if there is more than one address per fqdn: > [axels-macpro:~] axel% host cp4.lrau.net > cp4.lrau.net has address 91.216.35.77 > cp4.lrau.net has IPv6 address 2a02:d40:2:2::77 > [axels-macpro:~] axel% host 91.216.35.77 > 77.35.216.91.in-addr.arpa domain name pointer cp4.lrau.net. > [axels-macpro:~] axel% host 2a02:d40:2:2::77 > 7.7.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0.2.0.0.0.0.4.d.0.2.0.a.2.ip6.arpa domain name pointercp4.lrau.net. > [axels-macpro:~] axel% ping6 cp4 > PING6(56=40+8+8 bytes) 2a02:d40:2:10::121 --> 2a02:d40:2:2::77 > 16 bytes from 2a02:d40:2:2::77, icmp_seq=0 hlim=62 time=18.691 ms I don't believe we have any such check. What relevant configuration keys do you have set? -g

1 0

IPv6 support in 4.2?
by Axel Rau 21 Mar '13

21 Mar '13

I see some IPv6 patches in backport, while getting twisted.internet.error.CannotListenError: Couldn't listen on dead:beef:f:d::cc:8443: [Errno 43] Protocol not supported. Is this expected to work in 4.2? If yes, does the TLS/cert code handle dual stack addresses? Axel --- PGP-Key:29E99DD6 ☀ +49 151 2300 9283 ☀ computing @ chaos claudius

1 0

master process loops
by Axel Rau 18 Mar '13

18 Mar '13

While testing my new FreeBSD port on FreeBSD 8.2, I see the master process starting stealing cpu cycles after 1st client connect, so: PID JID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND 39240 11 root 3 76 0 126M 49784K uwait 6 1:38 17.77% python2.7 This continues after the last client has disconnected until shutdown of the server. I truss trace available on request. I see lots of these cycles: --- 39240: 569.691188580 gettimeofday({1363347523.922570 },0x0) = 0 (0x0) 39240: 569.691278526 gettimeofday({1363347523.922653 },0x0) = 0 (0x0) 39240: 569.691396825 gettimeofday({1363347523.922779 },0x0) = 0 (0x0) 39240: 570.519904713 kevent(3,{},0,{},25,{0.827203989 }) = 0 (0x0) 39240: 570.520012258 gettimeofday({1363347524.751392 },0x0) = 0 (0x0) 39240: 570.520110025 __sysctl(0x7fffffffbc40,0x2,0x7fffffffbbd0,0x7fffffffbc38,0x8064ba765,0xc) = 0 (0x0) 39240: 570.520148644 __sysctl(0x7fffffffbbd0,0x2,0x7fffffffbc90,0x7fffffffbcb8,0x0,0x0) = 0 (0x0) --- Axel --- PGP-Key:29E99DD6 ☀ +49 151 2300 9283 ☀ computing @ chaos claudius

3 2