I saw one report of a RAM issue.
I have 2G on the VM - and it was hung for > half a day. Another
report suggested a compiler bug, so I upgraded to the latest (gcc
5.1). I don't see the hang now, but errors such as --
/usr/include/boost/variant/variant.hpp:1575:38: error: call of
overloaded ‘initialize(void*, std::remove_reference<long long
int&>::type)’ is ambiguous
initializer::initialize(
^
The error output is attached - Please let me know if you can make sense of this.
--
Jai
On Tue, Mar 1, 2016 at 7:15 PM, Daniel Gryniewicz <dang@xxxxxxxxxx> wrote:
> On 02/29/2016 12:13 PM, Jai wrote:
>>
>> That was the output from make.
>> ps shows the actual compilation command as --
>>
>> jai 2822 2821 3 19:47 pts/0 00:05:09
>> /usr/libexec/gcc/x86_64-redhat-linux/4.8.5/cc1plus -quiet -I . -I
>> /usr/include/nss3 -I /usr/include/nspr4 -MD
>> test/encoding/ceph_dencoder-ceph_dencoder.d -MF
>> test/encoding/.deps/ceph_dencoder-ceph_dencoder.Tpo -MP -MT
>> test/encoding/ceph_dencoder-ceph_dencoder.o -D_GNU_SOURCE -D
>> HAVE_CONFIG_H -D __CEPH__ -D _FILE_OFFSET_BITS=64 -D _THREAD_SAFE -D
>> __STDC_FORMAT_MACROS -D _GNU_SOURCE -D CEPH_LIBDIR="/usr/local/lib" -D
>> CEPH_PKGLIBDIR="/usr/local/lib/ceph" -D GTEST_USE_OWN_TR1_TUPLE=0 -D
>> _REENTRANT -D WITH_RBD -D WITH_RADOSGW -U_FORTIFY_SOURCE
>> -D_FORTIFY_SOURCE=2 test/encoding/ceph_dencoder.cc -quiet -dumpbase
>> ceph_dencoder.cc -mtune=generic -march=x86-64 -auxbase-strip
>> test/encoding/ceph_dencoder-ceph_dencoder.o -g -g -O2 -O2
>> -Wtype-limits -Wignored-qualifiers -Winit-self -Wpointer-arith
>> -Wnon-virtual-dtor -Wno-invalid-offsetof -Wall -Wstrict-null-sentinel
>> -std=gnu++11 -fno-strict-aliasing -fsigned-char -ftemplate-depth=1024
>> -fexceptions -fPIE -fstack-protector -fno-builtin-malloc
>> -fno-builtin-calloc -fno-builtin-realloc -fno-builtin-free
>> -fno-var-tracking-assignments --param ssp-buffer-size=4 -o -
>>
>> It appears to be stuck in strcmp in libc.
>>
>> gstack 2822
>> #0 0x00007f57bd9b3006 in __strcmp_sse42 () from /lib64/libc.so.6
>> #1 0x000000000096447c in ?? ()
>> #2 0x0000000000dcd027 in htab_find_slot_with_hash ()
>> #3 0x000000000096b3b1 in ?? ()
>> #4 0x000000000098b741 in ?? ()
>> #5 0x0000000000e2f626 in ?? ()
>> #6 0x0000000000e3005c in toplev_main(int, char**) ()
>> #7 0x00007f57bd8a2b15 in __libc_start_main () from /lib64/libc.so.6
>> #8 0x0000000000dceaad in _start ()
>>
>> On Mon, Feb 29, 2016 at 8:45 PM, Loic Dachary <loic@xxxxxxxxxxx> wrote:
>>>
>>>
>>>
>>> On 29/02/2016 21:34, Jai wrote:
>>>>
>>>> Loic,
>>>>
>>>> With centos 7, I ve moved further .. The build is stuck however at this
>>>> --
>>>>
>>>> ./make_version -g ./.git_version
>>>> if [ -n "$NO_VERSION" ] ; then \
>>>> ./make_version -g ./.git_version -c ./ceph_ver.h -n ; \
>>>> else \
>>>> ./make_version -g ./.git_version -c ./ceph_ver.h ; \
>>>> fi
>>>> CXX test/encoding/ceph_dencoder-ceph_dencoder.o
>>>>
>>>> I ve tried restarting the make a few times after cleaning up, but it
>>>> comes back to this point. Could this be a failed/hung test ?
>>>
>>>
>>> I'm not sure, it looks like the compilation is not finished yet and the
>>> tests did not start. Could you provide details about what is stuck ?
>>>
>>> Cheers
>>>
>>>>
>>>> --
>>>> Jai
>>>>
>>>> On Sun, Feb 28, 2016 at 4:39 PM, Loic Dachary <loic@xxxxxxxxxxx> wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> Ubuntu 12.04 is no longer supported, reason why you're having troubles.
>>>>> I recommend CentOS 7 or Ubunut 14.04 instead.
>>>>>
>>>>> Cheers
>>>>>
>>>>> On 28/02/2016 17:55, Jai wrote:
>>>>>>
>>>>>> Loic,
>>>>>>
>>>>>> I did run install-deps before the make, and it completed without
>>>>>> error. I have an Ubuntu (1204) VM (uname shows 3.5.0-30-generic
>>>>>> #51~precise1-Ubuntu)
>>>>>>
>>>>>> I ran these as per the README --
>>>>>> ./install-deps.sh
>>>>>> ./autogen.sh
>>>>>> ./configure
>>>>>> make
>>>>>>
>>>>>> I found the bugs link on http://ceph.com/community/contribute/ under
>>>>>> "submit issues"
>>>>>>
>>>>>> --
>>>>>> Jai
>>>>>>
>>>>>> On Sun, Feb 28, 2016 at 4:18 PM, Loic Dachary <loic@xxxxxxxxxxx>
>>>>>> wrote:
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> You should try to run ./install-deps.sh first so that the required
>>>>>>> dependencies are all installed. Which operating system are you using ?
>>>>>>> The bug tracker is at http://tracker.ceph.com/. Where did you find a
>>>>>>> link to http://tracker.newdream.net/projects/ceph ? It should be fixed :-)
>>>>>>>
>>>>>>> Cheers
>>>>>>>
>>>>>>> On 28/02/2016 17:29, Jai wrote:
>>>>>>>>
>>>>>>>> I followed instructions in the README, but "make" gives me this
>>>>>>>> error --
>>>>>>>>
>>>>>>>> In file included from ./include/types.h:21:0,
>>>>>>>> from kv/LevelDBStore.h:6,
>>>>>>>> from kv/KeyValueDB.cc:5:
>>>>>>>> ./include/uuid.h:14:42: fatal error: boost/random/random_device.hpp:
>>>>>>>> No such file or directory
>>>>>>>> compilation terminated.
>>>>>>>>
>>>>>>>> I m compiling on ubuntu, and it looks like I have boost headers
>>>>>>>> (/usr/include/boost), but this file is absent. llibboost appears to
>>>>>>>> be
>>>>>>>> 1.46.1
>>>>>>>> appreciate any pointers to resolve this error.
>>>>>>>>
>>>>>>>> Also, I couldn't access the web page for bugs
>>>>>>>> (http://tracker.newdream.net/projects/ceph) - where can I find a
>>>>>>>> list
>>>>>>>> of bugs/ features ?
>>>>>>>> --
>>>>>>>> Jai
>
>
> Could you possibly not have enough RAM on the box? Ceph takes lots of RAM
> to build (I've run out with 4GB, and with 8G on a small parallel build). If
> you are heavily into swap, for example, individual compiles can take 10 of
> minutes each, and appear to be hung.
>
> Daniel
>
Attachment:
nohup.out
Description: Binary data
