Unable to start or initialise any ceph process (ceph-mon)

Bug #1629102 reported by James Page
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
ceph (Ubuntu)
Invalid
Undecided
Unassigned
google-perftools (Ubuntu)
Expired
Undecided
Unassigned
libunwind (Ubuntu)
Expired
High
Unassigned

Bug Description

Latest yakkety server build, with ceph 10.2.2; creation of the ceph-mon FS hangs:

ceph-mon --mkfs -i juju-3da7bf0b-0062-4a2f-8949-d85e6579b759-machine-15 --keyring /var/lib/ceph/tmp/juju-3da7bf0b-0062-4a2f-8949-d85e6579b759-machine-15.mon.keyring

spinning in:

nanosleep({0, 2000001}, NULL) = 0

top of backtrace:

#0 0x00007f5a619efd10 in __nanosleep_nocancel () at ../sysdeps/unix/syscall-template.S:84
No locals.
#1 0x00007f5a624b9193 in base::internal::SpinLockDelay(int volatile*, int, int) () from /usr/lib/libtcmalloc.so.4
No symbol table info available.
#2 0x00007f5a624b9026 in SpinLock::SlowLock() () from /usr/lib/libtcmalloc.so.4
No symbol table info available.
#3 0x00007f5a624adb78 in tcmalloc::ThreadCache::InitModule() () from /usr/lib/libtcmalloc.so.4
No symbol table info available.
#4 0x00007f5a624bbe55 in tc_malloc () from /usr/lib/libtcmalloc.so.4
No symbol table info available.
#5 0x00007f5a5fc05f7d in __fopen_internal (filename=0x7f5a6291c4a0 "/usr/lib/libtcmalloc.so.4", mode=0x7f5a5f984c6a "r", is32=1) at iofopen.c:69
        new_f = <optimized out>
#6 0x00007f5a5f982ab8 in ?? () from /usr/lib/x86_64-linux-gnu/libunwind.so.8
No symbol table info available.
#7 0x00007f5a5f983624 in ?? () from /usr/lib/x86_64-linux-gnu/libunwind.so.8
No symbol table info available.
#8 0x00007f5a5f983710 in _ULx86_64_dwarf_find_debug_frame () from /usr/lib/x86_64-linux-gnu/libunwind.so.8
No symbol table info available.
#9 0x00007f5a5f983cde in ?? () from /usr/lib/x86_64-linux-gnu/libunwind.so.8
No symbol table info available.
#10 0x00007f5a5fcdc0a4 in __GI___dl_iterate_phdr (callback=0x7f5a5f983940, data=0x7fff3a02ba90) at dl-iteratephdr.c:76
        __clframe = {__cancel_routine = <optimized out>, __cancel_arg = 0x0, __do_it = 1, __cancel_type = <optimized out>}
        nloaded = 26
        ns = <optimized out>
        caller = <optimized out>
        l = 0x7f5a6291c4c0
        info = {dlpi_addr = 140026172706816, dlpi_name = 0x7f5a6291c4a0 "/usr/lib/libtcmalloc.so.4", dlpi_phdr = 0x7f5a62489040, dlpi_phnum = 8,
          dlpi_adds = 26, dlpi_subs = 0, dlpi_tls_modid = 2, dlpi_tls_data = 0x7f5a62906760}
        ret = 0

looks like initial setup of tcmalloc is spinning - I've seen the same happen on start of the ceph-mon daemon (which I thought was the original problem).

ProblemType: Bug
DistroRelease: Ubuntu 16.10
Package: ceph-mon 10.2.2-0ubuntu5
ProcVersionSignature: Ubuntu 4.8.0-17.19-generic 4.8.0-rc7
Uname: Linux 4.8.0-17-generic x86_64
ApportVersion: 2.20.3-0ubuntu7
Architecture: amd64
Date: Thu Sep 29 20:43:51 2016
Ec2AMI: ami-0000045c
Ec2AMIManifest: FIXME
Ec2AvailabilityZone: nova
Ec2InstanceType: m1.small.osci
Ec2Kernel: unavailable
Ec2Ramdisk: unavailable
SourcePackage: ceph
UpgradeStatus: No upgrade log present (probably fresh install)

Revision history for this message
James Page (james-page) wrote :
summary: - Unable to start or initialise monitors
+ Unable to start or initialise any ceph process (ceph-mon)
Revision history for this message
James Page (james-page) wrote :

Reverting back to libunwind 1.1-4.1 resolves this problem for me - the changelog entry is:

libunwind (1.1-4.1ubuntu1) yakkety; urgency=medium

  * Configure with --enable-debug-frame. LP: #1590236.

 -- Matthias Klose <email address hidden> Mon, 19 Sep 2016 13:28:08 +0200

ceph was functional on yakkety not that long ago (this landed into updates on the 28th).

Revision history for this message
James Page (james-page) wrote :

I reverted that change and uploaded to:

 https://launchpad.net/~openstack-ubuntu-testing/+archive/ubuntu/ceph-sru

to test and confirm that appears to resolve this problem for me.

Changed in google-perftools (Ubuntu):
status: New → Invalid
Changed in libunwind (Ubuntu):
importance: Undecided → High
Changed in ceph (Ubuntu):
status: New → Invalid
Revision history for this message
Matthias Klose (doko) wrote :

does the issue persist when rebuilding ceph against the new libunwind?

Changed in ceph (Ubuntu):
status: Invalid → Incomplete
Revision history for this message
James Page (james-page) wrote :

No, the packages in:

 https://launchpad.net/~openstack-ubuntu-testing/+archive/ubuntu/ceph-sru/+build/10968224

where built using the 4.1ubuntu1 version from the archive and have the same symptom.

The dependency is via google-perftools rather that direct from ceph.

Changed in ceph (Ubuntu):
status: Incomplete → New
Revision history for this message
Matthias Klose (doko) wrote :

my bad, so maybe check with a rebuilt google-perftools?

Changed in google-perftools (Ubuntu):
status: Invalid → Incomplete
Revision history for this message
James Page (james-page) wrote :

Somehow this bug just went away - I'm not 100% sure why but marking invalid for ceph.

Changed in ceph (Ubuntu):
status: New → Invalid
Revision history for this message
Graham Inggs (ginggs) wrote :

libunwind (1.1-4.1ubuntu2) yakkety; urgency=medium

  * Don't configure with --enable-debug-frame on amd64.

 -- Matthias Klose <email address hidden> Fri, 30 Sep 2016 21:30:21 +0200

Changed in libunwind (Ubuntu):
status: New → Incomplete
Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for google-perftools (Ubuntu) because there has been no activity for 60 days.]

Changed in google-perftools (Ubuntu):
status: Incomplete → Expired
Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for libunwind (Ubuntu) because there has been no activity for 60 days.]

Changed in libunwind (Ubuntu):
status: Incomplete → Expired
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.