On 28-12-15 15:25, KeesBoog wrote:
> Thank you Florent and Wido,
>
> It seems that the sigsegv error occurs regardless of the code. Even if I
> do not upload a file, the error occurs after a couple of minutes...
> Strange!
>
Did you open a RADOS context or connection in that case?
What version of librados are you using on the system and what is the
Ceph cluster running?
Wido
> Thanks again,
>
> Kees
>
>
> Op 28-12-2015 om 13:54 schreef Florent Manens:
>> Hi Kees,
>>
>> I didn't test but try this :
>>
>> String file_id = "testupload_" + Id.getNext("file_id");
>>
>> try {
>> String config_file = "/etc/ceph/ceph.conf";
>> // nothing fancy here
>> String id = "admin";
>> String pool = "temp";
>>
>> Rados rados;
>> IoCTX ioctx;
>>
>> // make the connection
>> rados = new Rados(id) {
>> };
>> rados.confReadFile(new File(config_file));
>> rados.connect();
>> ioctx = rados.ioCtxCreate(pool);
>>
>> // read bytes from the inputstream
>> int read = 0;
>> long bytesRead = 0;
>> byte[] bytes = new byte[8192];
>>
>> while ((read = inputStream.read(bytes)) !=
>> -1) {
>> //truncate to really read data
>> byte[] real_buff = Arrays.copyOf(bytes, read);
>> ioctx.write(file_id, real_buff, bytesRead);
>> bytesRead += read;
>> }
>>
>> // close the connection
>> rados.ioCtxDestroy(ioctx);
>> rados.shutDown();
>> } catch (RadosException ex) {
>>
>> Logger.getLogger(ApiFunction_dvd_upload.class.getName()).log(Level.SEVERE,
>>
>> null, ex);
>> }
>>
>> I use Rados object for my Zimbra<->Ceph connector and I don't have
>> segfault problems with rados-java.
>>
>> Best regards,
>>
>>
>> ----- Le 28 Déc 15, à 13:43, KeesBoog techie2015@xxxxxxxxxxxxxx a écrit :
>>
>>> Hi Florent,
>>> Thanks.
>>> I did not consider 'plain Rados objects' and I am not sure how that
>>> would work. If you think that would solve the problem, perhaps you could
>>> elaborate?
>>> Would I still need Rbd to create a new image and set the size?
>>> In the example I replaced
>>> RbdImage image = rbd.open(file_id);
>>> image.write(bytes, bytesRead, read);
>>> with
>>> ioctx.write("rbd_id."+file_id, bytes, bytesRead);
>>> but
>>> a) the image-file was empty after the upload (so I am sure I am doing
>>> something wrong)
>>> b) the SIGSEGV-error was still there (after a few minutes)
>>> Best Regards,
>>> Kees
>>> Op 28-12-2015 om 12:33 schreef Florent Manens:
>>>> Hi Kees,
>>>> Why don't you use plain Rados objects instead of a RBD ?
>>>> Best regards,
>>>> ----- Le 28 Déc 15, à 11:54, KeesBoog techie2015@xxxxxxxxxxxxxx a
>>>> écrit :
>>>>> Thanks Wido,
>>>>> Here is the example code, taken from a Java web application.
>>>>> There are no other calls to Rbd/Rados in the code.
>>>>> // code altered for sake of the example
>>>>> (everything in one try/catch and random file name and initial size)
>>>>> // objective: read binary file(s) from an
>>>>> inputstream to a ceph image
>>>>> String file_id = "testupload_" +
>>>>> Id.getNext("file_id");
>>>>> long initialsize = 1000000000L;
>>>>> try {
>>>>> String config_file = "/etc/ceph/ceph.conf";
>>>>> // nothing fancy here
>>>>> String id = "admin";
>>>>> String pool = "temp";
>>>>> Rados rados;
>>>>> IoCTX ioctx;
>>>>> Rbd rbd;
>>>>> // make the connection
>>>>> rados = new Rados(id) {
>>>>> };
>>>>> rados.confReadFile(new File(config_file));
>>>>> rados.connect();
>>>>> ioctx = rados.ioCtxCreate(pool);
>>>>> rbd = new Rbd(ioctx);
>>>>> // create the file with initial size
>>>>> (larger than actual size)
>>>>> int features = (1 << 0);
>>>>> rbd.create(file_id, initialsize, features, 0);
>>>>> // open the image
>>>>> RbdImage image = rbd.open(file_id);;
>>>>> // read bytes from the inputstream
>>>>> int read = 0;
>>>>> long bytesRead = 0;
>>>>> byte[] bytes = new byte[8192];
>>>>> while ((read = inputStream.read(bytes)) !=
>>>>> -1) {
>>>>> image.write(bytes, bytesRead, read);
>>>>> bytesRead += read;
>>>>> }
>>>>> // adjust size to the actual read bytes
>>>>> image.resize(bytesRead);
>>>>> // close the image
>>>>> rbd.close(image);
>>>>> // close the connection
>>>>> rados.ioCtxDestroy(ioctx);
>>>>> rados.shutDown();
>>>>> } catch (RadosException ex) {
>>>>> Logger.getLogger(ApiFunction_dvd_upload.class.getName()).log(Level.SEVERE,
>>>>>
>>>>> null, ex);
>>>>> }
>>>>> This all works.
>>>>> I uploaded a very small text-file.
>>>>> rbd info temp/testupload_1028
>>>>> rbd image 'testupload_1028':
>>>>> size 5069 bytes in 1 objects
>>>>> order 22 (4096 kB objects)
>>>>> block_name_prefix: rbd_data.1f61b774b0dc51
>>>>> format: 2
>>>>> features: layering
>>>>> About 40 seconds after the successful upload (the content equals the
>>>>> original content) I get an unexpected SIGSEGV-error as reported
>>>>> (without
>>>>> a stacktrace).
>>>>> After a restart of Tomcat I did a second upload with a somewhat
>>>>> larger file:
>>>>> rbd info temp/testupload_1030
>>>>> rbd image 'testupload_1030':
>>>>> size 41151 kB in 11 objects
>>>>> order 22 (4096 kB objects)
>>>>> block_name_prefix: rbd_data.1f61c62ae8944a
>>>>> format: 2
>>>>> features: layering
>>>>> About 3 minutes after the successful upload I got the same
>>>>> SIGSEGV-error.
>>>>> And a third (first restart Tomcat again):
>>>>> rbd info temp/testupload_1032
>>>>> rbd image 'testupload_1032':
>>>>> size 127 MB in 32 objects
>>>>> order 22 (4096 kB objects)
>>>>> block_name_prefix: rbd_data.1f61fc2ae8944a
>>>>> format: 2
>>>>> features: layering
>>>>> About 2 minutes after the successful upload I got the same
>>>>> SIGSEGV-error.
>>>>> Thanks again.
>>>>> Kees
>>>>> Op 28-12-2015 om 10:21 schreef Wido den Hollander:
>>>>>> On 12/28/2015 07:18 AM, KeesBoog wrote:
>>>>>>> Hi,
>>>>>>> I do love rados-java! I do get listings of ceph pools; I even can
>>>>>>> upload
>>>>>>> files to a ceph pool with a Tomcat web application. That is great!
>>>>>> Great to hear!
>>>>>>> However, I get frequent SIGSEGV errors (which halt Tomcat):
>>>>>> Ok, that is not good. Can you reproduce it with maybe some example
>>>>>> code?
>>>>>> I would like to dig down to where this is happening, but if you
>>>>>> are able
>>>>>> to reproduce it with some simple example code it would be great.
>>>>>> Or a stacktrace on which call it actually dies.
>>>>>> Wido
>>>>>>> #
>>>>>>> # A fatal error has been detected by the Java Runtime Environment:
>>>>>>> #
>>>>>>> # SIGSEGV (0xb) at pc=0x00007fd4bc54166d, pid=25308,
>>>>>>> tid=140554357282560
>>>>>>> #
>>>>>>> # JRE version: OpenJDK Runtime Environment (7.0_91-b02) (build
>>>>>>> 1.7.0_91-b02)
>>>>>>> # Java VM: OpenJDK 64-Bit Server VM (24.91-b01 mixed mode
>>>>>>> linux-amd64
>>>>>>> compressed oops)
>>>>>>> # Derivative: IcedTea 2.6.3
>>>>>>> # Distribution: Ubuntu 14.04 LTS, package
>>>>>>> 7u91-2.6.3-0ubuntu0.14.04.1
>>>>>>> # Problematic frame:
>>>>>>> # C [librados.so.2+0x28766d] Mutex::Lock(bool)+0xd
>>>>>>> #
>>>>>>> # Failed to write core dump. Core dumps have been disabled. To
>>>>>>> enable
>>>>>>> core dumping, try "ulimit -c unlimited" before starting Java again
>>>>>>> #
>>>>>>> # An error report file with more information is saved as:
>>>>>>> # /tmp/hs_err_pid25308.log
>>>>>>> #
>>>>>>> # If you would like to submit a bug report, please include
>>>>>>> # instructions on how to reproduce the bug and visit:
>>>>>>> # http://icedtea.classpath.org/bugzilla
>>>>>>> #
>>>>>>> Please find the file hs_err_pid25308.log attached. I could send a
>>>>>>> few more.
>>>>>>> It would be great if someone could point me to a workaround of
>>>>>>> perhaps
>>>>>>> even fix the issue.
>>>>>>> Best Regards,
>>>>>>> Kees
>>
>>> _______________________________________________
>>> ceph-users mailing list
>>> ceph-users@xxxxxxxxxxxxxx
>>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
