1250813901 Q * Piet Remote host closed the connection
1250815030 J * Piet ~piet@659AABYGK.tor-irc.dnsbl.oftc.net
1250817109 J * scientes_ ~scientes@97-113-166-198.tukw.qwest.net
1250817525 Q * scientes__ Ping timeout: 480 seconds
1250817740 Q * docelic Quit: http://www.spinlocksolutions.com/
1250818808 J * derjohn_foo ~aj@f049133065.adsl.alicedsl.de
1250819258 Q * derjohn_mob Ping timeout: 480 seconds
1250819993 J * biz_ biz@baze.de
1250819993 Q * biz Read error: Connection reset by peer
1250823263 Q * Piet Ping timeout: 480 seconds
1250823973 J * Piet ~piet@659AABYIJ.tor-irc.dnsbl.oftc.net
1250824831 Q * _gh_ Ping timeout: 480 seconds
1250826483 J * scientes__ ~scientes@97-113-167-117.tukw.qwest.net
1250826881 Q * scientes_ Ping timeout: 480 seconds
1250827383 Q * geb Ping timeout: 480 seconds
1250829994 J * doener ~doener@i59F5495F.versanet.de
1250830097 Q * doener_ Ping timeout: 480 seconds
1250830368 Q * ensc Ping timeout: 480 seconds
1250830885 J * ensc ~irc-ensc@77.235.160.26
1250831738 Q * Piet Ping timeout: 480 seconds
1250831865 N * ensc Guest295
1250831865 Q * Guest295 Remote host closed the connection
1250831875 J * ensc ~irc-ensc@77.235.160.26
1250831950 J * Piet ~piet@659AABYJX.tor-irc.dnsbl.oftc.net
1250832910 J * balbir_ ~balbir@122.172.9.7
1250832948 Q * derjohn_foo Ping timeout: 480 seconds
1250833519 J * sharkjaw ~gab@149-49-95.oke2-bras6.adsl.tele2.no
1250833645 Q * nkukard Ping timeout: 480 seconds
1250833669 J * ghislainocfs2 ~Ghislain@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr
1250834569 Q * scientes__ Quit: scientes__
1250835593 J * ghislainocfs21 ~Ghislain@adsl2.aqueos.com
1250835628 J * mxs mxs@p4FCCA9EF.dip.t-dialin.net
1250835776 J * derjohn_foo ~aj@195.243.216.22
1250835942 Q * mxs_ Ping timeout: 480 seconds
1250835950 Q * ghislainocfs2 Ping timeout: 480 seconds
1250836313 J * ghislainocfs2 ~Ghislain@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr
1250836645 J * nkukard ~nkukard@196.212.73.74
1250836677 Q * ghislainocfs21 Ping timeout: 480 seconds
1250837019 Q * balbir_ Read error: Connection reset by peer
1250837316 M * derjohn_foo Bertl_zZ, http://paste.linux-vserver.org/13393
1250837361 M * derjohn_foo Bertl_zZ, the wonderbra exploit did work in my guest (devel branch), but there might be missing some stuff (it wants to execute mplayer )
1250837381 Q * sharkjaw Remote host closed the connection
1250840144 N * Bertl_zZ Bertl
1250840148 M * Bertl morning folks!
1250840164 M * Bertl derjohn_foo: you mean, did _not_ work, no?
1250840233 M * derjohn_foo Bertl, correct, it did _not_ work opposed to the guy who reported here, it would work.
1250840271 M * derjohn_foo It looks you were right ... at least in a "plain case" (i.e. ipx and ppp were not loaded on the server etc.)
1250840333 M * derjohn_foo I wonder why mainline didnt fix the 2.6.29-branch ... at least I cannot find the entry in the changelog. time to move to 2.6.30... 
1250840390 M * derjohn_foo hm, patch-2.6.30.5-vs2.3.0.36.14-pre5.diff   <--- "pre" sound like "fragile, do not touch!
1250840702 M * Bertl it's not complete, but it works quite fine ... I use it on some of my servers
1250841119 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net
1250842461 M * derjohn_foo not complete means features are missing or new kernel featutes of 2.6.30 are not adopted yet ?
1250842536 M * Bertl features are missing
1250842545 M * derjohn_foo btw: is there any progress in "suspending/freezing" a vs-guest?
1250842587 M * Bertl you can 'pause' guests with the TB extension
1250842600 M * derjohn_foo TB ? Thunderbird *g* ? 
1250842617 M * Bertl Token Bucket (term used since 3 years :)
1250842650 M * derjohn_foo why is it am extension? I know the TB only as a scheduler?
1250842654 M * derjohn_foo *an
1250842683 M * derjohn_foo so, simply give it "zero buckets" and it is somewhat freezed ...
1250842701 M * derjohn_foo but cant be migrated to a differenz machine with that
1250842828 M * Bertl the TB scheduler extension, extends the scheduler :)
1250842871 M * Bertl and there is a 'pause' flag, and no, it cannot be migrated (as single guest), mainline is not that ready yet
1250842898 M * Bertl but you didn't ask for migration either :)
1250842932 M * Bertl you can still use kvm or xen to 'migrate' blocks or the entire system, in case of maintainance
1250843299 M * derjohn_foo well, be currently deploy xen-domUs (in PV mode) with vserver enabled kernels within the domU
1250843803 M * Bertl k, off for now ... got some work to do in the basement
1250843871 N * Bertl Bertl_oO
1250843914 J * kir ~kir@swsoft-msk-nat.sw.ru
1250843979 J * rozzzmarin 5f9ee8a4@webchat.mibbit.com
1250844047 M * rozzzmarin Hi, I'm newbie to IRC. I can post my question here?
1250844244 M * rozzzmarin I want to install Zimbra mail server on my debian vserver. Zimbra requires 127.0.0.1. I read that all services binding to localhost will bind to interface/0. But if I will use interface/0 as RFC 1918 and interface/1 as public ip i don't have internet inside my vserver 
1250844260 M * rozzzmarin Anybody can help me to solve this problem? Thanks in advance
1250844669 M * Bertl_oO rozzzmarin: depends on you kernel version and guest setup
1250844728 M * rozzzmarin My kernel is 2.6.18-6-vserver-686
1250844745 M * Bertl_oO so, an older debian kernel ...
1250844747 M * Bertl_oO http://linux-vserver.org/Problematic_Programs#Zimbra_Mail
1250844765 M * Bertl_oO note that with a recent kernel, you have full lo isolation, so nothing special needs to be done
1250844820 M * rozzzmarin so, the best solution is to install the recent kernel?
1250844837 M * rozzzmarin and don't have troubles with lo anymore?
1250844881 M * Bertl_oO probably, but note that the 2.6.26 debian kernel is broken, so you do not want to use that
1250844918 M * Bertl_oO mainline 2.6.27.x should be fine though
1250844951 M * rozzzmarin thanks!
1250845013 M * Bertl_oO np
1250846746 J * Kmos ~gothicx@kmos.user.oftc.net
1250847195 J * gnuk ~F404ror@pla93-3-82-240-11-251.fbx.proxad.net
1250847378 M * arekm hm, some bug in .27 vserver patches again? 54567 bind(4, {sa_family=AF_INET6, sin6_port=htons(80), inet_pton(AF_INET6, "::", &sin6_addr), sin6_flowinfo=0, sin6_scope_id=0}, 28) = -1 EADDRINUSE (Address already in use)
1250847428 M * arekm netstats shows no binded services on 80 in guest and on host
1250847531 M * arekm and I can successfuly bind in this guest IF I stop apache from other guest
1250847698 J * balbir_ ~balbir@122.172.5.230
1250847709 M * arekm the funny thing is that if I stop apache in other (2) guest, then start in 1 guest and then I'll still be able to start in 2 guest
1250847833 M * arekm er, this is of course bug in 2.6.30.x patch, not 27
1250848056 N * DoberMann[ZZZzzz] DoberMann
1250851676 Q * balbir_ Ping timeout: 480 seconds
1250851676 Q * Kmos Read error: Connection reset by peer
1250852333 J * balbir_ ~balbir@122.172.46.214
1250853803 Q * rozzzmarin Quit: http://www.mibbit.com ajax IRC Client
1250855095 Q * Evet_ Read error: Connection reset by peer
1250856586 J * vServer_User ~vServer_U@host90-152-0-28.ipv4.regusnet.com
1250856861 Q * vServer_User 
1250860169 J * bakins ~bakins@157.166.167.129
1250860217 M * bakins has anybody looked at what it would take to get vserver to work with eucalyptus?  performance of Xen and KVM has just been dreadful in my testing...
1250861732 N * groente_ groente
1250861983 M * _Shiva_ bakins: eucalyptus supports OpenVZ - i think adding vServer should at least be possible...
1250862010 M * bakins _Shiva_: is openvz only in dev versions?  does it use libvrit? 
1250862502 M * _Shiva_ hmm - it uses libvirt from at least 1.4rc ... so it _should_ support any virtualization/isolation supported by libvirt..
1250862524 M * _Shiva_ but it states only Xen and KVM in it's docs..
1250862547 M * _Shiva_ never tried it, though ;-)
1250862563 M * _Shiva_ . o 0 ( yet.. it's on my agenda .. )
1250865316 Q * derjohn_foo Ping timeout: 481 seconds
1250865993 J * micah_ ~micah@micah.riseup.net
1250865999 J * mugwump_ ~samv@watts.utsl.gen.nz
1250866004 J * tam_ ~tam@gw.nettam.com
1250866007 J * mstrobert_ ~mstrobert@wkstn.wycliffe.ca
1250866022 J * Supaplex_ ~supaplex@166.70.62.193
1250866044 M * Guy- I'm trying to run virtualbox-3.0 inside a vserver guest, and an ioctl on /dev/vboxdrv is returning EINVAL
1250866049 M * Guy- could this be caused by vserver?
1250866102 Q * mugwump resistance.oftc.net larich.oftc.net
1250866102 Q * mstrobert resistance.oftc.net larich.oftc.net
1250866102 Q * tam resistance.oftc.net larich.oftc.net
1250866102 Q * Supaplex resistance.oftc.net larich.oftc.net
1250866102 Q * micah resistance.oftc.net larich.oftc.net
1250866102 Q * AndrewLee resistance.oftc.net larich.oftc.net
1250866306 J * AndrewLee ~andrew@u7.hlc.edu.tw
1250867209 Q * balbir_ Read error: Connection reset by peer
1250867440 M * Guy- well, it _is_ caused by vserver, because it works on the host...
1250867638 M * Guy- also, what I can do about it? I tried giving the guest all bcaps, but that didn't help
1250867722 Q * Pazzo Quit: .
1250868043 Q * tokkee Ping timeout: 480 seconds
1250868202 M * daniel_hozac is /dev/vboxdrv the same on the host and the guest?
1250868831 J * geb ~geb@AOrleans-253-1-1-145.w90-24.abo.wanadoo.fr
1250869020 J * balbir_ ~balbir@122.172.157.242
1250869243 N * DoberMann DoberMann[PullA]
1250869598 J * tokkee tokkee@osprey.tokkee.org
1250869602 Q * FireEgl Read error: Connection reset by peer
1250869795 J * larsivi ~larsivi@70.84-48-63.nextgentel.com
1250870101 Q * tokkee Ping timeout: 480 seconds
1250870268 J * tokkee tokkee@osprey.tokkee.org
1250870587 J * FireEgl Proteus@WTF.4.1.0.c.0.7.4.0.1.0.0.2.ip6.arpa
1250870860 Q * MooingLemur Quit: leaving
1250870894 J * MooingLemur ~troy@shells195.pinchaser.com
1250870896 Q * MooingLemur 
1250870907 J * MooingLemur ~troy@shells195.pinchaser.com
1250871066 Q * tokkee Ping timeout: 480 seconds
1250871210 J * tokkee tokkee@osprey.tokkee.org
1250872529 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d
1250873269 J * geb_ ~geb@earth.gebura.eu.org
1250873563 Q * geb Ping timeout: 480 seconds
1250875201 Q * ghislainocfs2 Quit: Leaving.
1250875283 J * ghislainocfs2 ~Ghislain@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr
1250875513 N * micah_ micah
1250876612 Q * nenolod Quit: my eyes cannot compute this misery.
1250876622 J * nenolod nenolod@petrie.dereferenced.org
1250877108 J * derjohn_foo ~aj@e180193190.adsl.alicedsl.de
1250877212 N * DoberMann[PullA] DoberMann
1250877560 J * mib_ej6yg1yf bc249fe7@webchat.mibbit.com
1250877568 P * mib_ej6yg1yf 
1250879159 N * Bertl_oO Bertl
1250879164 M * Bertl back now ...
1250879432 Q * gnuk Quit: NoFeature
1250879892 P * bakins 
1250880087 Q * FireEgl Remote host closed the connection
1250880236 Q * balbir_ Ping timeout: 480 seconds
1250880487 J * hparker ~hparker@2001:470:1f0f:32c:290:96ff:fe50:40fa
1250880943 J * balbir_ ~balbir@122.172.31.112
1250880948 J * FireEgl FireEgl@173-16-9-10.client.mchsi.com
1250881373 M * arekm Bertl: enter /last arekm for bugreport
1250881445 M * Bertl what util-vserver version?
1250881493 M * Bertl and does the guest use ipv6 only, or ipv6 and ipv4?
1250881495 M * arekm 0.30.216-0.pre2841
1250881517 M * arekm one uses ipv4 and ipv6, the other one just ipv4
1250881532 M * Bertl can you reproduce the issue with ipv4 only?
1250881658 M * arekm binding to 0.0.0.0 succeeded in the same setup. binding to :: fails
1250881698 M * Bertl so it is definitely ipv6 related, okay, could you upload the virtnet/* info for both guests, and tell me which one fails?
1250881758 M * arekm hm, now why I don't have /proc/virtnet...
1250881767 M * Bertl on the host
1250881773 M * arekm riight
1250881868 M * arekm oh crap
1250881945 M * arekm I lied a bit. both have ipv6 enabled: http://pld.pastebin.com/f67057bd2
1250882069 M * arekm note that I have few more guests on this machine and no idea why guest 700 had impact on guest 100 and not the other guests
1250882469 M * Bertl hmm, that's indeed strange ...
1250882718 M * Bertl could you try with just the ipv4 addresses and once just with the ipv6 addresses?
1250882816 M * arekm I'll do that but that won't be now (need to setup separate test env for that)
1250883332 M * Bertl okay, I'll look over the code if I see something obvious ... maybe bonbons finds a minute to look at it too ...
1250883662 M * bonbons arekm: how is your IPv6 socket on the failing side?
1250883680 M * bonbons is it IPv6 only or does it allow mapped IPv4?
1250883699 M * arekm bonbons: bind(4, {sa_family=AF_INET6, sin6_port=htons(80), inet_pton(AF_INET6, "::", &sin6_addr), sin6_flowinfo=0, sin6_scope_id=0}, 28) = -1 EADDRINUSE (Address already in use)
1250883739 M * arekm net.ipv6.bindv6only is 0
1250883762 M * arekm and there is no v6only sock opt set
1250883950 M * bonbons ok, so if I understand you case correctly, service of guest 700 listening on 0.0.0.0:X prevents service of guest 100 to listen on [::]:X (and both guests don't share any IP address)
1250884065 M * arekm almost, looks like 700 is listening on 193.239.45.149:80 not 0.0.0.0:80 (according to netstat -nlp)
1250884092 M * arekm and yes, 100 is trying to bind to [::]:80 which fails while 0.0.0.0:80 works
1250884138 M * bonbons listening on 193.239.45.149:80 can be caused by single-address optimization, so not really important :)
1250884145 M * bonbons what kerne patch are you running?
1250884174 M * arekm patch-2.6.30.5-vs2.3.0.36.14-pre5.diff
1250884194 M * bonbons thanks, will look at it
1250884620 M * Bertl hmm, shouldn't the single ip be off with 2 IPs (regardless whether they are ipv4 or ipv6)?
1250884695 M * Bertl arekm: you mentioned 2.6.27.x before, does it work on 2.6.27.x or do you see similar there?
1250884723 M * bonbons don't know... short test here, one guest has IPv4 only, one has IPv6 and IPv6, IPv4 guest listens on 0.0.0.0:10000 and mixed guest can listen to [::]:10000 without problem (2.6.29-vs2.3.0.36-9-pre3)
1250884740 M * bonbons testing with netcat
1250884757 M * Bertl well, could be something we broke in 2.6.30
1250884804 M * bonbons yep, next step is installing & compiling 2.6.30-vs$arekm and checking there
1250884831 M * bonbons but easierest is to check first with most recent vs kernel installed on test box!
1250884906 M * Bertl hmm, checking my list, I indeed see some ipv6 todo points
1250884928 M * Bertl i.e. we had some rejects, and I have to check if they were resolved
1250885439 M * Bertl yep, the patch to net/ipv6/addrconf.c is missing, as the ipv6_rcv_saddr_equal() was removed, probably moved to net/ipv6/udp.c?
1250885488 M * Bertl and a hunk to net/ipv6/af_inet6.c (checking for ipv4)
1250886013 M * bonbons Bertl: see here for mainline change: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=b2f5e7cd3dee2ed721bf0675e1a1ddebb849aee6
1250886249 M * Bertl ah, yes, I see ... well, do you want to look into it?
1250886387 M * bonbons I can look into it, though first need to understand the new version of ipv6_rcv_saddr_equal()
1250886500 M * bonbons (and I don't know why it was moved to udp! (no reason in commit message, and extern remains in addrconf header, except maybe for matching IPv4 layout)
1250886579 M * Bertl no idea here either .. do you like me to upload the rejected hunks?
1250886681 M * bonbons that would make life easier, e.g. no need to extract it from patch for 2.6.29
1250886693 M * Bertl okay, give me a second ...
1250886877 M * Bertl http://vserver.13thfloor.at/ExperimentalT/addrconf.c.rej
1250886889 M * Bertl http://vserver.13thfloor.at/ExperimentalT/af_inet6.c.rej
1250887061 M * fb good evening!
1250887157 M * Bertl evening!
1250887267 Q * ghislainocfs2 Quit: Leaving.
1250889329 J * ViRUS ~mp@p57A6EB1F.dip.t-dialin.net
1250889421 M * ViRUS Why haven't there any stable releases for the last two years? Are the "experimental" releases to be considered stable, or is the code really not suitable for production use?
1250889471 M * Bertl there are no 'recent' stable releases, because the 'users' do not support the 'developers' to do the code review and testing
1250889505 M * Bertl the "experimental" releases are "experimental" but they are also used in production by many folks
1250889533 M * FIChTe we run the latest pre patch for kernel 2.6.30, no crashes or strange behavior so far
1250889553 M * FIChTe running 40+ vservers 32bit and 64bit on a dual quadcore xeon machine
1250889556 M * FIChTe heavy loaded
1250889567 M * FIChTe distcc´s for 32 and 64bit
1250889588 M * Bertl  but, as you see above, ipv6 is broken (just an example)
1250889595 M * ViRUS I see. So basicly it's just a naming thing. As in several projects. It's just a naming thing. Like "unstable" debian is not really unstable. Even though strange things can happen sometimes. ;)
1250889597 M * FIChTe yes
1250889709 M * FIChTe bertl whats the important things to test? i could do this part
1250889737 M * FIChTe testing all posix caps in any combinations for example?
1250889761 M * Bertl http://linux-vserver.org/Feature_Matrix
1250889782 M * Bertl basically all listed features need to be tested
1250889946 M * biz_ Humm.. a regression test suite would be nice?
1250889978 M * Bertl sure, go ahead, write one ... testme and testfs are a good start, I guess
1250890002 M * Bertl I had plans to do automated testing with qemu/kvm on several architectures
1250890040 M * Bertl but as user contributions seized about a year ago (except for notable exceptions), there was no time to spend on that
1250890141 Q * saulus Ping timeout: 480 seconds
1250890171 J * uva_ bno@118-160-168-185.dynamic.hinet.net
1250890341 J * saulus ~saulus@c152174.adsl.hansenet.de
1250890415 M * ViRUS mmh, I'd love to contribute, but to be true I'm just a vserver user myself and haven't ever done any kernel coding at all. Besides time is short and I prefer spending time on application development. ;)
1250890556 Q * uva Ping timeout: 480 seconds
1250890831 Q * derjohn_foo Ping timeout: 480 seconds
1250890857 M * Marillion harry: sorry for noise, but is it plan from 2.6.27.15 to forward the grsec patch to expand 2.6.27.27?
1250891237 M * biz_ doener: I just saw your entry on the Developers page.. are you the doener who's active in Freenode's #git? ;-)
1250892229 M * biz_ Bertl: Such a testsuite like a "ready to test linux-vserver on your architecture qemu/kvm image" would be really nice. I guess it would yield more test-reports by users.. the effort (preparation to be able to run the tests, cleaning up etc.) involved with your testme.sh and testfs.sh hinders that at least for me
1250892333 M * Bertl well, if you want to design something like that, I can probably give you a few hints
1250892583 M * daniel_hozac the preparation required is in the 3 command area.
1250892807 M * bonbons Bertl: do you know by heart if addr_in_nx_info() is NULL-safe for the address (NULL == any)?
1250892922 M * Bertl you mean, if it works for 0.0.0.0 as address or with a missing nxinfo?
1250892991 M * biz_ daniel_hozac: I doubt that? For example, I don't have a free test-system nor any of my desktop machines run vserver-patched kernels
1250892994 M * bonbons in mean if e.g. for ipv6 struct in6_addr* arg is NULL
1250893032 M * Bertl ah, v4 uses a direct argument, for v6, I don't know
1250893061 M * bonbons ok, then I will read code towards leaf to find out
1250893094 M * biz_ daniel_hozac: and the vserver systems I'm running are running the old stable with grsec since they were released.. in production. Not available for testing either :(
1250893094 M * Bertl but as it uses the mainline functions, I'd expect that to work
1250893095 M * bonbons quite some work to merge, especially attempting to fix the FIXME at same time ;)
1250893130 M * Bertl nope, ipv6_masked_addr_cmp for example cannot handle NULL
1250893156 M * Bertl same for ipv6_addr_equal, so no, NULL will caus a panic
1250893236 M * bonbons ok, then I will have to be really careful... as in ipv6_rcv_saddr_equal() sk2's saddr may be NULL (at least according to the code in there!)
1250893389 M * bonbons daniel_hozac: is there still a tool in util-vserver that runs a command in a (new) network context?
1250893401 M * Bertl ncontext, chbind
1250893441 M * bonbons can it configure the context on the fly? e.g. ncontext $addr-lst -- $cmd
1250893470 M * daniel_hozac chbind creates the context and adds the addresses and sets the attributes.
1250893478 M * daniel_hozac ncontext just creates the context.
1250893486 M * Bertl otherwise use ncontext, naddress
1250893496 M * bonbons so that the context goes aways as soon as $cmd (and sub-processes if any) exit
1250893514 M * Bertl or the sockets have been closed, yes
1250893537 M * bonbons well, the idea is to be able to create a bash script for testing the address collision detection
1250893546 M * Bertl (i.e. all holders are gone)
1250893653 Q * ViRUS Quit: ViRUS
1250893675 M * bonbons if someone wans to write such a (partial) test-suite for the network contexts, go ahead and tell me when there is something to use :)
1250893716 M * Bertl testnet.sh or testnx.sh :)
1250893730 M * Bertl but yeah, needs to be written :)
1250893816 M * bonbons would be damn handy to validate the IPv6 implementation of saddr_equal!
1250894052 N * DoberMann DoberMann[ZZZzzz]
1250895085 M * bonbons hmm, af_inet6.c in inet6_bind() looks like in IPv4-mapped err is not being set before goto out ...
1250895125 M * bonbons need to check when (and what commit message says) that thing was changed in mainline as it looks broken!
1250895154 M * Bertl it might be the result of a partial apply, if you are looking at the Linux-VServer patch
1250895214 M * bonbons could be too, though according to your reject and the code block that looks less probable, but I've already spotted the probable changeset, so matter of looking at it
1250895255 M * bonbons it's here: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=63d9950b08184e6531adceb65f64b429909cc101
1250895329 M * bonbons so the question is whether IPv4 also just returns 0 in that case (might be a bind to 0.0.0.0 in which case it could be fine)
1250895384 M * bonbons though that looks rather improbable (that it would be bind to 0.0.0.0)
1250895571 M * bonbons yeah, that's a bug, IPv4 first sets err to -EADDRNOTAVAIL before doing the test which does the goto
1250898956 Q * hparker Quit: Read error: 104 (Peer reset by connection)
1250899174 Q * dowdle Remote host closed the connection