[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
I am closing this because the bug has been fixed in Maverick and Natty (10.10 and onwards). If anyone still needs a fix for the bug in previous versions of Ubuntu, please do steps 1 and 2 of the SRU Procedure [1] to bring the need to a developer's attention. [1]: https://wiki.ubuntu.com/StableReleaseUpdates#Procedure ** Changed in: libpcap (Ubuntu) Status: Confirmed = Fix Released -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/414744 Title: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/libpcap/+bug/414744/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
Does anybody know if there's a ppa which has fixes for this I can use in Lucid? My SIP servers run tcpdump constantly and we get an occasion about once a week one one server or another where this is affecting service. Cheers, Neil. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/414744 Title: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
Launchpad provides a convenient method of searching. Just go to https://launchpad.net/ubuntu/+source/libpcap and scroll down and click on the Other versions of 'libpcap' in untrusted archives. and you'll get a list of all the PPAs with libpcap. I honestly don't know if any of the listed PPAs include this patch, but the patch and the source are available, so it would probably be fairly straightforward to roll your own. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/414744 Title: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
... this will be fixed in a lucid update. Um, not exactly. And I'm not particularly eager to update to Maverick, either. (And why the frack did they call it Maverick? Do they want me to think of John effin McCain every time I update packages?) -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
Unfortunately, I was unable to get permission for a Stable Release Update on this, so it doesn't look like it will be going into Lucid. I really don't know why, and I'm rather peeved about it as well. Really can't comment on the naming conventions, I'm just another user. -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
I was lucky to see an update of the description these days (even if it mostly changed tabs to spaces or so). Let me congratulate to Ubuntu Lucid Lynx 10.04 LTS (s stands for support, eh?). Ok, Jaunty - no support. Karmic - no support. With Lucid, we've at least one thing: longer time for finishing the support. But in the change of the description we read: + The issue was fixed upstream as part of version 1.1.0 (now 1.1.1), which will be in Maverick as soon as Debian imports begin. + The attached patch, while developed independently, is effectively identical to the fix implemented upstream. - No support for LongTermSupport Lucid(!). It will be fixed in the short-time-supported successor Maverick. Furhermore we read that someone else has developed independently a patch. That's nice. But even if 10 or 100 different people would have invested their spare time to develop seperatly the same patch (what a non-sense ;) - it wouldn't have helped to get it into ubuntu. I'm really, really upset. A high-CPU-load problem with a patch available needs one year for still- not-being resolved. How on the other hand will it be with a security issue (segfaults and so on), when no one provides a patch (or no one submits it to launchpad, because we've learned it does not help)? Time to leave. This has no future. bye -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
The tabs/spaces thing was an accident, you can ignore them. If you'll look under the bug status, you'll see Nominated for Lucid. That basically means that this will be fixed in a lucid update after it has gotten some testing (probably in a few weeks). The full upstream version will be in Maverick, but most of the bugfixes it contains are in the process of being applied to the version in Lucid. The full new version is not in Lucid because it contains a few untested new features that weren't suitable for an LTS. -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
** Description changed:
+ Libpcap goes into an infinite spin, using 100% of CPU when the interface it
is watching is brought down.
+ The issue was fixed upstream as part of version 1.1.0 (now 1.1.1), which will
be in Maverick as soon as Debian imports begin.
+ The attached patch, while developed independently, is effectively identical
to the fix implemented upstream.
+
+ TEST CASE:
+ - run tcpdump on a network interface (ex: sudo tcpdump -i eth0)
+ - bring down the interface (ex: sudo ifconfig eth0 down)
+ - watch as tcpdump uses 100% cpu
+
+ Original Report:
Binary package hint: libpcap0.8
Programs using libpcap0.8, like tcpdump, show the behavior that they consume
100% CPU power instead of error-handling / termination.
It's a problem in libpcap. I traced it down and present a fix here.
I wrote to ubuntu-devel-disc...@lists.ubuntu.com
(https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-August/009194.html)
and the maintainer, Romain Francoise, on Aug 7. But there was no response.
This fix goes belong with my report
https://bugs.launchpad.net/ubuntu/+source/tcpdump/+bug/369288 (concerning
tcpdump), which I have closed because I found it's a libpcap problem.
Perhaps it may be needed to check why it occured: has libpcap recently
changed in the poll() handling - or did the linux kernel counterpart change?
test screnario:
r...@tomate:/tmp# openvpn --dev tun
r...@tomate:/tmp# ifconfig tun0 up
r...@tomate:/tmp# tcpdump -ni tun0
-
strace -tt -p 12238 /tmp/a 21
10:14:39.250055 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
10:14:40.251505 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
10:14:41.252910 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
10:14:42.254355 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
10:14:43.255764 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
10:14:44.257102 poll([{fd=3, events=POLLIN}], 1, 1000) = 0 (Timeout)
here i terminated openvpn; dev tun0 closes
10:14:45.258137 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.392988 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393089 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393182 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393257 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393361 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393436 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393520 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393621 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393694 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393851 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.393924 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.394020 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.394093 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.394165 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:45.394260 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
[..]
10:14:58.921514 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
10:14:58.921558 poll([{fd=3, events=POLLIN}], 1, 1000) = 1 ([{fd=3,
revents=POLLERR}])
-
tho...@tomate:~$ grep ^10:14:45 /tmp/a|wc
- 12508 112572 1088196
+ 12508 112572 1088196
tho...@tomate:~$ grep ^10:14:46 /tmp/a|wc
- 22874 205866 1990038
+ 22874 205866 1990038
tho...@tomate:~$ grep ^10:14:47 /tmp/a|wc
- 22488 202392 1956456
+ 22488 202392 1956456
tho...@tomate:~$ grep ^10:14:48 /tmp/a|wc
- 22853 205677 1988211
+ 22853 205677 1988211
tho...@tomate:~$ grep ^10:14:49 /tmp/a|wc
- 22924 206316 1994388
+ 22924 206316 1994388
tho...@tomate:~$ grep ^10:14:50 /tmp/a|wc
- 22975 206775 1998825
+ 22975 206775 1998825
tho...@tomate:~$ grep ^10:14:51 /tmp/a|wc
- 22130 199170 1925310
+ 22130 199170 1925310
tho...@tomate:~$ grep ^10:14:52 /tmp/a|wc
- 22778 205002 1981686
+ 22778 205002 1981686
tho...@tomate:~$ grep ^10:14:53 /tmp/a|wc
- 23061 207549 2006307
+ 23061 207549 2006307
tho...@tomate:~$ grep ^10:14:54 /tmp/a|wc
- 22982 206838 1999434
+ 22982 206838 1999434
tho...@tomate:~$ grep ^10:14:55 /tmp/a|wc
- 22937 206433 1995519
+ 22937 206433 1995519
tho...@tomate:~$ grep ^10:14:56 /tmp/a|wc
- 22906 206154 1992822
+ 22906 206154 1992822
tho...@tomate:~$ grep ^10:14:57 /tmp/a|wc
- 22742 204678
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
** Package changed: libpcap0.8 (Ubuntu) = libpcap (Ubuntu) -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
On 2010-03-13 21:30:38 -, eapache eapa...@gmail.com wrote in 20100313213038.2388.27067.mal...@soybean.canonical.com: I'm so sorry! I've been triaging and working on bugs filed against the libpcap package for a while now, and I only just realized that there are also bugs filed against the libpcap0.8 package. I'm really not happy with this. This bug appeared in Ubuntu Jaunty and I submitted it a few days after it's release (2009-04-29). It was not fixed and I did not even get a response. A few days after Karmic's release (2009-08-17) where the bug was still present I developed and submitted a patch. Again nothing happened (no reaction, no thanks, no technical critics on the patch) - until 2010-03-13. And now you say: I've added your patch as an attachment. I've also been in contact with upstream, and there is a new upstream version coming down the pipe in the next short while. I don't think there's much hope for karmic anymore, but I should be able to get this into Lucid one way or another. - no hope for Karmic? That means two complete Ubuntu versions with this bug. Karmic is supposed to be supported until April 2011 (- +1 year from now). We are not discussing about a new feature or a new library version here. It's a bugfix which could be added - quickly, after a code review. Please note that Jaunty is also affected, which is supported until October 2010 - and where it will never be fixed. In the past, new Ubuntu versions always introduced new bugs. I.e. practical inusability of X11 on my graphic-chip on Jaunty which was stable over years with previous Ubuntu versions. It was (partly) fixed by advertising to upgrade to the release candidate of karmic; see https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/ 419958 [partly means: no crashes anymore (previously 2 days); but in karmic, sometimes characters are not display until you scroll or mark it with your mouse..] Or look at report https://bugzilla.gnome.org/show_bug.cgi?id=575548 (gnome-session-manager bug, on Intrepid); it was fixed with the upgrade to Jaunty. But: Intrepid is still supported (until April 2010). With Karmic, I've noted down about 35 new bugs (some of them were fixed over the time, some goes daily on my nerves). With a response time of 0.5-2 years to a bug submitted on launchpad and the perspective to get a fix for the next or next-to-next Ubuntu version it's not really motivating to invest the time to submit all the bugs I discovered. Instead, it may be recommended to switch over to another linux distribution - something I tried to avoid over the years, regardless how unhappy I was. Ubuntu has real, heavy problems. Because you as a maintainer are in a better position to report the disappointments of the users about the quality issues upstream, I've wrote my complains in detail here. I've to stress that I do not blame you. It's also ok to oversee a bug report. But due to my experience, it's a common problem ubuntu has - and they have to solve this quality problem, otherwise users will leave. I've the ear on the local community, and I know I'm not alone with my concerns. Kind regards, - Thomas Osterried ** Bug watch added: GNOME Bug Tracker #575548 https://bugzilla.gnome.org/show_bug.cgi?id=575548 -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
I'm really not happy with this. Believe me, I feel your pain. A lot of the problem is manpower - Ubuntu has over 25,000 packages all told, and we have nowhere close to that many people working on bugs. I've subscribed to the libpcap packages now, so in future any bugs filed against libpcap, libpcap0.8, etc. should expect a prompt and hopefully helpful response from me in a day or two. However, these are just a few packages in the archive. As our user-base grows, I expect (and hope) that relatively major packages like this one will be picked up, but it won't happen overnight. no hope for Karmic? That means two complete Ubuntu versions with this bug. Karmic is supposed to be supported until April 2011 (- +1 year from now). We are not discussing about a new feature or a new library version here. It's a bugfix which could be added - quickly, after a code review. Please note that Jaunty is also affected, which is supported until October 2010 - and where it will never be fixed. This has been a point of long-standing debate. Basically what it boils down to is Ubuntu's official StableReleaseUpdate policy https://wiki.ubuntu.com/StableReleaseUpdates. Because Ubuntu is used in enterprise environments, the powers that be have decided the bug you know is better than the one you don't. I don't necessarily always agree, but that is the policy I have to follow. With a response time of 0.5-2 years to a bug submitted As I mentioned, it really depends if the package has any dedicated subscribers. If it does then the bug should be handled promptly. If it doesn't, then reporting bugs against that package is almost (but not quite) useless. There are a few brave souls who triage bugs against all packages, but they simply can't keep up with the volume, so bugs get missed. In the past, new Ubuntu versions always introduced new bugs. Yes. Regressions are terrible. Again, a lot of it comes down to manpower. We could only completely avoid regressions if we never pulled in new upstream versions of software. At that point we basically become Debian. It is my experience that the new system for handling regressions put in place around Jaunty's release helped a lot for Karmic (although your experience seems to disagree), but this is also something that should improve over time as we get more users. Ubuntu has real, heavy problems. Because you as a maintainer are in a better position to report the disappointments of the users about the quality issues upstream, I've wrote my complains in detail here. I've to stress that I do not blame you. It's also ok to oversee a bug report. But due to my experience, it's a common problem ubuntu has. I'm not a maintainer :) I've subscribed to the libpcap packages so I can see bug reports as they come in, but I have no upload privileges or any other official permissions. I do agree with you, but I don't see an easy solution - the people who manage bugs for Ubuntu only have so many hours in their lives, and it just isn't enough. If you do want to bring this conversation to the attention of people higher-up, I suggest sending it to the Ubuntu-devel-discuss mailinglist at https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss, which is a public list for exactly this kind of issue to be raised. There's always a couple of core devs hanging around there. I hope I've answered your questions. Sincerely, Evan Huus -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
ah, ok. and now? I'm not sure what you're asking. There really isn't much for us to do until the new libpcap makes it into Debian. Then I can file a sync request, get it in the repositiories, and make sure that this bug has been fixed. -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
I'm so sorry! I've been triaging and working on bugs filed against the libpcap package for a while now, and I only just realized that there are also bugs filed against the libpcap0.8 package. I've added your patch as an attachment. I've also been in contact with upstream, and there is a new upstream version coming down the pipe in the next short while. I don't think there's much hope for karmic anymore, but I should be able to get this into Lucid one way or another. Thanks, Evan ** Patch added: patch.diff http://launchpadlibrarian.net/40913969/patch.diff ** Changed in: libpcap0.8 (Ubuntu) Status: New = Confirmed -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
** Tags added: patch -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
Re: [Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
On 2010-03-13 22:31:07 -, Brian Murray br...@ubuntu.com wrote in 20100313223108.6527.61193.launch...@soybean.canonical.com: ** Tags added: patch ah, ok. and now? -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
Anyone here? - It can't be so difficult to apply the patch I submitted above - four months ago. -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
[Bug 414744] Re: libpcap issue: tcpdump consumes 100% cpu power after interface shutdown
This bug still exists in karmic. Btw, it's a bit disappointing. I provided a fix here - and there was no response, no action, anything. -- libpcap issue: tcpdump consumes 100% cpu power after interface shutdown https://bugs.launchpad.net/bugs/414744 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
