0

I've got a CentOS 6.5 x86-64 KVM server with a bunch of guest VMs of different breeds, mostly EL5 and EL6. However one and only one of them keeps crashing every couple of days with:

pthread_create failed: Resource temporarily unavailable

Here is the full log from /var/log/libvirt/qemu/vws3-pp.log:

2014-07-24 21:27:27.451+0000: starting up
LC_ALL=C PATH=/sbin:/usr/sbin:/bin:/usr/bin QEMU_AUDIO_DRV=none
    /usr/libexec/qemu-kvm -name vws3-pp,process=qemu:vws3-pp -S -M rhel6.5.0
    -enable-kvm -m 1536 -redhat-disable-KSM -realtime mlock=on
    -smp 1,sockets=1,cores=1,threads=1 -uuid d11de823-8bab-4e8d-8457-61ef7ab877a7
    -nodefconfig -nodefaults -chardev socket,id=charmonitor,path=/var/lib/libvirt/qemu/vws3-pp.monitor,server,nowait
    -mon chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown
    -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2
    -drive file=/vm/prod/vws3-pp-disk1.qcow2,if=none,id=drive-virtio-disk0,format=qcow2,cache=writethrough
    -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x4,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1
    -netdev tap,fd=22,id=hostnet0,vhost=on,vhostfd=32
    -device virtio-net-pci,netdev=hostnet0,id=net0,bus=pci.0,addr=0x3
    -chardev pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0
    -vnc 127.0.0.1:9,password -vga cirrus
    -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5
char device redirected to /dev/pts/5

pthread_create failed: Resource temporarily unavailable   <====  ### HERE ####
2014-07-29 15:29:52.063+0000: shutting down

There are 8 other VMs on the box and all of them run happily for months, just this one crashes every few days. There is nothing special about this VM - pretty standard LAMP, not overloaded - I can't think of any significant difference between this and the other VMs that exhibit no problems. Some of those are very busy but still rock stable.

Somewhere on the net I found a suggestion to set max_processes = 4096 in /etc/libvirt/qemu.conf and restart the box - done that but it didn't help. The VM crashed again this morning for no good reason.

NEW INFO:

As it turns out the VM always dies while rdiff-backup is running from a remote backup server and in most cases the last log in the rdiff-backup-data/backup.log (on the remote side, ie not affected by the crash) is:

Processing changed file tmp
Incrementing mirror file /extpool/backup/vws3-pp/tmp

Even though /tmp/** is excluded from the backup. It could indeed be failing in /usr which is the next one alphabetically in /, who knows...

Backup runs every night but the VM crashes only about once a week.

What does rdiff-backup do so strange that it makes a KVM gues die with pthread_create failed: Resource temporarily unavailable?

Any ideas?

1 Answer 1

0

Check

cat /proc/`pidof qemu-kvm`/limits

to see if the limits are actually set after restart.

1
  • Yes it's set: Max processes 4096 4096 processes
    – MLu
    Jul 31, 2014 at 4:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .