Bug#591286: libapr1: upgrade breaks apache
On Sun, Aug 1, 2010 at 22:49:00 +0200 (+0200), Stefan Fritsch wrote: [snip] My sincere apologies for the delay in replying - I thought I had but I see now that I didn't. A great bit of work there tracking down the problem and workarounds :-) > Out of interest: Have you noticed any performance difference with > sslmutex sem instead of the default 'file'? The description of the > types on http://httpd.apache.org/docs/2.2/misc/perf-tuning.html says > that the semaphore mutex types are somewhat less robust. I've not noticed any problems (nor measured any performance increases, if any). The posixsem type doesn't seem _too_ bad, the sysvmem one certainly does. Possibly a little experiment if I find myself bored one weekend :-) Cheers, Adrian -- bitcube.co.uk - Expert Linux infrastructure consultancy Puppet, Debian, Red Hat, Ubuntu, CentOS -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#591286: libapr1: upgrade breaks apache
On Sunday 01 August 2010, Adrian Bridgett wrote: > i386 (but it's a KVM guest, sorry - I should have said). I found out why posixsem does not work: A bug in pbuilder/cowbuilder causes posix shared mem/posix semaphores to not work in the build chroot and this causes apr's configure to disable it. Since I always use cowbuilder to build the packages, all i386 libapr1 packages of the last few years had non-working posixsem. The only exception was 1.2.12-5+lenny1 which was not uploaded by me but by Peter Samuelson. You were simply lucky that you upgraded to lenny after version 1.2.12-5+lenny1 was released. With 1.2.12-5 it would not have worked either. If you need it, you can fix it by rebuilding libapr1. It should work if you either don't use pbuilder, or use pbuilder and put /dev/shm in the BINDMOUNTS variable in /etc/pbuilderrc. I haven't yet looked into the issue of "sem" trying to use posixsem even if it is not available. This really does look like a bug in apr or apache. Out of interest: Have you noticed any performance difference with sslmutex sem instead of the default 'file'? The description of the types on http://httpd.apache.org/docs/2.2/misc/perf-tuning.html says that the semaphore mutex types are somewhat less robust. Cheers, Stefan -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#591286: libapr1: upgrade breaks apache
On Sun, Aug 1, 2010 at 20:51:19 +0200 (+0200), Stefan Fritsch wrote: > On Sunday 01 August 2010, Adrian Bridgett wrote: > > However, if I downgrade _just_ libapr1 to 1.2.12-5+lenny1 then > > posixsem (and sem) work just fine. > > Which architecture are you using? Wow, that's a fast response! :-) i386 (but it's a KVM guest, sorry - I should have said). host and guest are both on 2.6.34-rc6 kernels. I'm using qemu-kvm 0.12.3. Guest is stared via libvirt: 29457 ?Sl 12126:49 /usr/bin/kvm -S -M pc-0.12 -enable-kvm -m 750 -smp 1,sockets=1,cores=1,threads=1 -name bishop -uuid bf7828e4-dfe1-ed21-7eef-ef820b8f6d48 -nodefaults -chardev socket,id=monitor,path=/var/lib/libvirt/qemu/bishop.monitor,server,nowait -mon chardev=monitor,mode=readline -boot c -drive file=/dev/rootvg/bishop-disk,if=none,id=drive-virtio-disk0,boot=on,cache=writeback -device virtio-blk-pci,bus=pci.0,addr=0x4,drive=drive-virtio-disk0,id=virtio-disk0 -drive file=/dev/rootvg/bishop-backup,if=none,id=drive-virtio-disk1,cache=writeback -device virtio-blk-pci,bus=pci.0,addr=0x5,drive=drive-virtio-disk1,id=virtio-disk1 -drive file=/dev/rootvg/mp3lv,if=none,id=drive-virtio-disk2,cache=writeback -device virtio-blk-pci,bus=pci.0,addr=0x6,drive=drive-virtio-disk2,id=virtio-disk2 -device virtio-net-pci,vlan=0,id=net0,mac=00:16:3e:01:04:00,bus=pci.0,addr=0x7 -net tap,fd=28,vlan=0,name=hostnet0 -chardev pty,id=serial0 -device isa-serial,chardev=serial0 -usb -device usb-tablet,id=input0 -vnc 0.0.0.0:3 -k en-gb -vga cirrus -device usb-host,hostbus=001,hostaddr=002,id=hostdev0 -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x3 Adrian -- bitcube.co.uk - Expert Linux infrastructure consultancy Puppet, Debian, Red Hat, Ubuntu, CentOS -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#591286: libapr1: upgrade breaks apache
On Sunday 01 August 2010, Adrian Bridgett wrote: > However, if I downgrade _just_ libapr1 to 1.2.12-5+lenny1 then > posixsem (and sem) work just fine. Which architecture are you using? -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#591286: libapr1: upgrade breaks apache
Package: libapr1 Version: 1.2.12-5+lenny2 Severity: grave I used to have "SSLMutex sem" however testing an upgrade today this breaks with: [Sun Aug 01 19:09:36 2010] [error] (70023)This function has not been implemented on this platform: Cannot create SSLMutex Configuration Failed The apache mod_ssl manual says that it should "pick the best semaphore implementation available". Trying posixsem fails but sysvsem works so it should be using that (another bug really). However, if I downgrade _just_ libapr1 to 1.2.12-5+lenny1 then posixsem (and sem) work just fine. Looking at the changelog, this could be nasty :-( Thanks, Adrian -- bitcube.co.uk - Expert Linux infrastructure consultancy Puppet, Debian, Red Hat, Ubuntu, CentOS -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
