Summary: | net-dns/avahi-0.6.31-r8: avahi-daemon constantly reporting "Invalid response packet from host" | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Pacho Ramos <pacho> |
Component: | Current packages | Assignee: | Anthony Basile <blueness> |
Status: | RESOLVED FIXED | ||
Severity: | normal | ||
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | http://avahi.org/ticket/352 | ||
See Also: |
https://launchpad.net/bugs/1342400 https://bugzilla.redhat.com/show_bug.cgi?id=1240711 |
||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Pacho Ramos
2015-09-02 11:25:56 UTC
(In reply to Pacho Ramos from comment #0) > I am getting lots of messages like: > avahi-daemon[401]: Invalid response packet from host > > People in https://bugs.launchpad.net/ubuntu/+source/avahi/+bug/1342400 > wonder if maybe this is caused by some new Windows 10 or latest Mac OS X > computers running around, but looks like they still couldn't find a real fix > for this :S > http://avahi.org/ticket/352 I'm not sure there's anything for me to do. All indications is that its an macosx thing. avahi drops these packets so its safe. I will try to follow the bug in Ubuntu side (as upstream looks to not care for now) as this even causes avahi to use 1-2% CPU only spamming journalctl with tons of lines like this :( (In reply to Pacho Ramos from comment #2) > I will try to follow the bug in Ubuntu side (as upstream looks to not care > for now) as this even causes avahi to use 1-2% CPU only spamming journalctl > with tons of lines like this :( okay so you're seeing this. what i can do is mute those warnings or add some code to give the warning only once every X seconds if repeated if i can safely schedule an alarm signal, or once every X messages if i can't. do you think this will help with the cpu usage? Maybe it will help :/ As a side note, this is the old patch that worked some ago: http://lists.freedesktop.org/archives/avahi/attachments/20120719/1e71846e/attachment.obj But per Ubuntu latest comment, it looks to not be enough now :S Do you know if there is any way to simply blacklist and IP from avahi? I can use that as a workaround because the machines that are causing this warnings are not going to be used by me ever ;) Thanks Fedora has applied this for this bug: http://pkgs.fedoraproject.org/cgit/avahi.git/commit/?id=54e63ffbd52feaf8acf71ab5b0480dc14ba4dbce There are also some other fixes pushed recently in Fedora for other bugs that maybe could be useful too :) (In reply to Pacho Ramos from comment #6) > Fedora has applied this for this bug: > http://pkgs.fedoraproject.org/cgit/avahi.git/commit/ > ?id=54e63ffbd52feaf8acf71ab5b0480dc14ba4dbce > > There are also some other fixes pushed recently in Fedora for other bugs > that maybe could be useful too :) have you tested it and does it fix the problem? I will try the next Monday (as that is the network that has the ton of systems causing this warnings) (In reply to Pacho Ramos from comment #8) > I will try the next Monday (as that is the network that has the ton of > systems causing this warnings) I did a build test with that patch and everything seems to be okay. The avahi ebuild has epatch_user so its convenient to test patches. Just remove the avahi.spec chunk since its just to build rpms. the patch is not enough :'( I will reply in redhat bug report :( Fedora updated the patch and finally it works :D http://pkgs.fedoraproject.org/cgit/avahi.git/plain/avahi-0.6.31-invalid_packet.patch (In reply to Pacho Ramos from comment #11) > Fedora updated the patch and finally it works :D > http://pkgs.fedoraproject.org/cgit/avahi.git/plain/avahi-0.6.31- > invalid_packet.patch please test avahi-0.6.31-r9.ebuild which adds this patch. if it works we can stabilize since its a small change over -r8 which is currently stable. Please note r9 ebuild has a typo that ends up with the patch not being applied eptch "${FILESDIR}"/${P}-invalid_packet.patch eptch instead of epatch :) Also maybe for -r9 and fast stabilization this is great, but I would also opt for a -r10 with some additional patches coming from upstream git that would probably be interesting (as I wonder when 0.6.32 would be released), they are being applied for a long time on Fedora without issues. I could maybe prepare that ebuild, introduce it hardmasked to let you review it if desired and unmask it after that. Anyway, I plan to mostly follow the patches applied in Fedora (In reply to Pacho Ramos from comment #13) > Please note r9 ebuild has a typo that ends up with the patch not being > applied > eptch "${FILESDIR}"/${P}-invalid_packet.patch > > eptch instead of epatch :) Once it's changed to epatch it works fine ;) (In reply to Pacho Ramos from comment #13) > Please note r9 ebuild has a typo that ends up with the patch not being > applied > eptch "${FILESDIR}"/${P}-invalid_packet.patch > > eptch instead of epatch :) fix in -r10. r9 is now off the tree. sorry about that. reopen if this is still a problem. test for about a week and we'll reapid stab. |