Development for LHC@home

This new Native version for Linux should work even if CVMFS has not been installed.

Note that it isn't working yet and I am cycling through versions.

Now on v4.35. Should be working.

Hi Laurence,
have only booted a SL610 VM without any corrections:
https://lhcathomedev.cern.ch/lhcathome-dev/workunit.php?wuid=1907493

Hi Laurence,
have only booted a SL610 VM without any corrections:
https://lhcathomedev.cern.ch/lhcathome-dev/workunit.php?wuid=1907493

It won't work on SL6 as we need a later kernel for rootless containers.

Here is a successful task.

Here is a CentOS7 Task.
https://lhcathomedev.cern.ch/lhcathome-dev/results.php?hostid=3830
Had first forgotten the namespaces command:
echo 640 > /proc/sys/user/max_user/namespaces

After setting this command got this Error:
<core_client_version>7.14.2</core_client_version>
<![CDATA[
<message>
process exited with code 195 (0xc3, -61)</message>
<stderr_txt>
21:03:27 (15625): wrapper (7.15.26016): starting
21:03:27 (15625): wrapper (7.15.26016): starting
21:03:27 (15625): wrapper: running ../../projects/lhcathomedev.cern.ch_lhcathome-dev/cranky-0.1.1 ()
21:03:27 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Detected Theory App
21:03:27 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Checking CVMFS.
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Checking runc.
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Creating the filesystem.
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Using /cvmfs/cernvm-prod.cern.ch/cvm3
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Updating config.json.
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO] Running Container 'runc'.
nsenter: failed to unshare namespaces: Operation not permitted
container_linux.go:336: starting container process caused "process_linux.go:279: running exec setns process for init caused \"exit status 46\""
21:03:49 CEST +02:00 2019-07-11: cranky-0.1.1: [ERROR] Container 'runc' terminated with status code 1.
21:03:50 (15625): cranky exited; CPU time 0.514570
21:03:50 (15625): app exit status: 0xce
21:03:50 (15625): called boinc_finish(195)
21:03:50 CEST +02:00 2019-07-11: cranky-0.1.1: [INFO]
</stderr_txt>
]]>

cvmfs_config probe
atlas.cern.ch, atlas-condb.cern.ch, grid.cern.ch, cernvm-prod.cern.ch, sft.cern.ch, alice.cern.ch are all ok!

SL76 first Task successful.
Only echo 640... was needed to start in Console.
https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2789523

1. CVMFS Deinstallation over Software in SL76.

2. In Console sudo yum install -y cvmfs

cvmfs.x86_64 0:2.6.0-1.el7 successful installed.
3. Task https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2790793
have this error
<stderr_txt>
08:30:48 (3842): wrapper (7.15.26016): starting
08:30:48 (3842): wrapper (7.15.26016): starting
08:30:48 (3842): wrapper: running ../../projects/lhcathomedev.cern.ch_lhcathome-dev/cranky-0.1.1 ()
08:30:48 CEST +02:00 2019-07-14: cranky-0.1.1: [INFO] Detected Theory App
08:30:48 CEST +02:00 2019-07-14: cranky-0.1.1: [INFO] Checking CVMFS.
cvmfs2: /home/sftnight/jenkins/workspace/CvmfsFullBuildDocker/CVMFS_BUILD_ARCH/docker-x86_64/CVMFS_BUILD_PLATFORM/cc7/build/BUILD/cvmfs-2.6.0/cvmfs/util/posix.cc:409: void loader::WritePipe(int, const void*, size_t): Assertion `(num_bytes >= 0) && (static_cast<size_t>(num_bytes) == nbyte)' failed.
cvmfs2: /home/sftnight/jenkins/workspace/CvmfsFullBuildDocker/CVMFS_BUILD_ARCH/docker-x86_64/CVMFS_BUILD_PLATFORM/cc7/build/BUILD/cvmfs-2.6.0/cvmfs/util/posix.cc:409: void loader::WritePipe(int, const void*, size_t): Assertion `(num_bytes >= 0) && (static_cast<size_t>(num_bytes) == nbyte)' failed.
cvmfs2: /home/sftnight/jenkins/workspace/CvmfsFullBuildDocker/CVMFS_BUILD_ARCH/docker-x86_64/CVMFS_BUILD_PLATFORM/cc7/build/BUILD/cvmfs-2.6.0/cvmfs/util/posix.cc:409: void loader::WritePipe(int, const void*, size_t): Assertion `(num_bytes >= 0) && (static_cast<size_t>(num_bytes) == nbyte)' failed.
08:30:49 CEST +02:00 2019-07-14: cranky-0.1.1: [WARNING] 'cvmfs_config probe sft.cern.ch' failed.
08:30:49 CEST +02:00 2019-07-14: cranky-0.1.1: [INFO] Creating local CVMFS repository.
./cvmfs-mini-0.1-amd64/mount_cvmfs.sh: Zeile 33: 3941 Speicherzugriffsfehler (Speicherabzug geschrieben) LD_LIBRARY_PATH="${script_location}/lib" CVMFS_LIBRARY_PATH="${script_location}/bin" ${script_location}/bin/cvmfs2 -o "config=${script_location}/default.conf:${script_location}/default.local,libfuse=2" "$REPOSITORY" "$MOUNTPOINT"
08:30:49 CEST +02:00 2019-07-14: cranky-0.1.1: [WARNING] 'cvmfs_config probe sft.cern.ch' failed.
08:30:49 (3842): cranky exited; CPU time 0.411743
08:30:49 (3842): app exit status: 0xce
08:30:49 (3842): called boinc_finish(195)

Installation of cvmfs come from Docker and why is it cvmfs-2.6.0 ?
Will test it again with a reboot after deinstallation of cvmfs.

1. CVMFS Deinstallation over Software in SL76.

2. In Console sudo yum install -y cvmfs

cvmfs.x86_64 0:2.6.0-1.el7 successful installed.

Will test it again with a reboot after deinstallation of cvmfs.

After reboot and echo... it is now running.

Running Debian 10 4.19.0-5-amd64 without CVMFS installed.

https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2791669

stderr:

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<message>
process exited with code 195 (0xc3, -61)</message>
<stderr_txt>
22:16:04 (9820): wrapper (7.15.26016): starting
22:16:04 (9820): wrapper (7.15.26016): starting
22:16:04 (9820): wrapper: running ../../projects/lhcathomedev.cern.ch_lhcathome-dev/cranky-0.1.1 ()
22:16:04 CEST +02:00 2019-07-19: cranky-0.1.1: [INFO] Detected Theory App
22:16:04 CEST +02:00 2019-07-19: cranky-0.1.1: [INFO] Checking CVMFS.
22:16:04 CEST +02:00 2019-07-19: cranky-0.1.1: [WARNING] 'which' could not locate the command 'cvmfs_config'.
22:16:04 CEST +02:00 2019-07-19: cranky-0.1.1: [INFO] Creating local CVMFS repository.
Segmentation fault
22:16:04 CEST +02:00 2019-07-19: cranky-0.1.1: [WARNING] 'cvmfs_config  sft.cern.ch' failed.
22:16:05 (9820): cranky exited; CPU time 0.196851
22:16:05 (9820): app exit status: 0xce
22:16:05 (9820): called boinc_finish(195)
</stderr_txt>
]]>

In dmesg:

[126929.347969] cvmfs2[9847]: segfault at 0 ip 0000000000000000 sp 00007ffffb2da988 error 14 in cvmfs2[557df2b4e000+5000]
[126929.347972] Code: Bad RIP value.

It's a quite plain install, only enabled user namespaces.

user@computer:~$ cat /proc/sys/user/max_user_namespaces
63225

Did I miss something?

/dda

No you didn't miss anything. The CVMFS code we are using does not work on all Linux flavours. We need to rebuild to make it supports more platforms.

Have tested -dev native and Production -native in ONE Boinc under Linux SL76.

In -dev there is always creating a new slot-number when a new task is running.
The old slot is empty after task is finished, but a new slot-number is taken.

When slot number 100 is reached than Boinc send a message and no new task is running.

3 User and three unsuccessful tasks:
https://lhcathomedev.cern.ch/lhcathome-dev/workunit.php?wuid=1910459

Seems to be quite a few of these 195 errors where the Tasks fail around the same point (90mins or so) for all 3 users.