Event Builder Servers 2025
List of available servers
All machines are Supermicro servers.
hostnames (gsi) |
hostnames (hades) |
OS |
cores |
memory |
disk /data01 |
disk other |
purpose |
remarks |
lxhadesdaq |
lxhadesdaqp |
Deb11 5.10.226-1 (2024-10-03) |
32 (AMD EPYC 7351P) |
64G |
- |
7.3 T drbd + 440G /sdai |
central server nfs/dhcp/dns; Oracle client gateway |
hardware of two machines lxhadesdaq1/2 with shared drbd |
lxhadeb08 |
lxhadeb08p |
Deb11 5.10.221-1 (2024-07-14)) |
32 (AMD EPYC 7351) |
252G |
73 T |
44G /sdai |
BNET eb 1/5/9/13 |
|
lxhadeb10 |
lxhadeb10p |
Deb11 5.10.223-1 (2024-08-10) |
32 (AMD EPYC 7351) |
252G |
73 T |
386G /sdai |
BNET eb 3/7/11/15 |
also hosts /glusterfs/lgad_georep |
lxhadeb17o |
lxhadeb17po |
Deb11 5.10.237-1 (2025-05-19) |
48 (AMD EPYC 7443P) |
252G |
64 T |
443G /dev |
BNET input 4 |
additional fast optical nic with different hostname |
lxhadeb18o |
lxhadeb18po |
Deb11 5.10.237-1 (2025-05-19) |
48 (AMD EPYC 7443P) |
252G |
64 T |
443G /dev |
BNET input 5 |
additional fast optical nic with different hostname |
lxhadeb14 |
lxhadeb14p |
Deb11 5.10.226-1 (2024-10-03) |
64 (AMD EPYC 7302) |
252G |
51 T |
224G /sdbi |
BNET input 1 |
also hosts /glusterfs/lgad_georep |
lxhadeb15 |
lxhadeb15p |
Deb11 5.10.223-1 (2024-08-10) |
64 (AMD EPYC 7302) |
252G |
51 T |
224G /sdai |
BNET input 2 |
also hosts /glusterfs/lgad_georep |
lxhadeb16 |
lxhadeb16p |
Deb11 5.10.226-1 (2024-10-03) |
64 (AMD EPYC 7302) |
252G |
51 T |
224G /sdbi |
BNET input 3 |
also hosts /glusterfs/lgad_georep |
lxhadeb09 |
lxhadeb09p |
Deb11 5.10.223-1 (2024-08-10) |
32 (AMD EPYC 7351) |
252G |
45 T |
32G /sdbi |
BNET eb 2/6/10/14 |
|
lxhadeb11 |
lxhadeb11p |
Deb11 5.10.223-1 (2024-08-10) |
32 (AMD EPYC 7351) |
252G |
44 T |
20G /sdai |
BNET eb 4/8/12/16 |
|
lxhadeb13 |
lxhadeb13p |
Deb12 6.1.140-1 (2025-05-22) |
64 (AMD EPYC 7351) |
252G |
22 T |
346G /sdbi |
spare for BNET master and raw monitor |
|
lxhadeb12 |
lxhadeb12p |
Deb11 5.10.223-1 (2024-08-10) |
32 (AMD EPYC 7351) |
252G |
7.3 T |
25G /sdai |
BNET master, dabc raw monitor server |
|
--
Main.JoernAdamczewski - 2025-09-26 - 08:22
Software components to install
example: set up server lxhadeb13 with all required components
lxhadeb13
with Debian GNU/Linux 12 (bookworm), gcc version 12.2.0 (Debian 12.2.0-14+deb12u1), cmake version 3.25.1
system packages
These are missing on a blank
lxhadeb*
machine- TODO: put already in chef role
- apt-get install aptitude
- aptitude install autoconf
- aptitude install cmake
- aptitude install gcc
- aptitude install g++
- aptitude install gpp
- aptitude install automake
- aptitude install libtool
- aptitude install libz-dev
- aptitude install libattr1-dev
- aptitude install uuid-dev
- aptitude install make
LTSM and FSQ libraries
Sources and all components under
/home/hadaq/ltsm/
-
/home/hadaq/ltsm/ibm_tsm
: ibm tsm system packages (required for ltsm only, not necessarily for fsq). Subfolder with version 8.1.15
- dpkg -i gskcrypt64_8.0-55.24.linux.x86_64.deb
- dpkg -i gskssl64_8.0-55.24.linux.x86_64.deb
- dpkg -i tivsm-api64.amd64.deb
- dpkg -i tivsm-apicit.amd64.deb
- dpkg -i tivsm-ba.amd64.deb
- dpkg -i tivsm-bacit.amd64.deb
- dpkg -i gskssl64_8.0-55.24.linux.x86_64.deb
-
/home/hadaq/ltsm/certs
: certificate that must be registered in system by using:
-
dsmcert -add -server ltsmi1 -file /home/hadaq/ltsm/certs/cert256-ltsmi1-25092023.cert
- NOTE: dsmcert is part of the above packages, but installation does not provide all libs in search path!
- had to make some soft links manually:
- ln -s /usr/local/ibm/gsk8_64/lib64/libgsk8ssl_64.so /lib
- ln -s /usr/local/ibm/gsk8_64/lib64/libgsk8iccs_64.so /lib
- ln -s /usr/local/ibm/gsk8_64/lib64/libgsk8km_64.so /lib
- ln -s /usr/local/ibm/gsk8_64/lib64/libgsk8cms_64.so /lib
-
/home/hadaq/ltsm/install
: the installation directory (redirects to install2025
)
-
/home/hadaq/ltsm/ltsm
: source of ltsm lib git clone of https://github.com/GSI-HPC/ltsm
- build with
./autogen.sh
-
./configure --prefix="/home/hadaq/ltsm/install2025" CFLAGS='-g -DDEBUG -O0' --enable-test
-
make
; make install
- manually provide link for binaries in system path at
/usr/local/bin/ltsmc
to /home/hadaq/ltsm/install2025/bin/ltsmc
-
/home/hadaq/ltsm/fsq
: source of fsq lib git clone of https://github.com/GSI-HPC/fsq
- build with
./autogen.sh
-
./configure --prefix="/home/hadaq/ltsm/install2025" CFLAGS='-g -DDEBUG -O0' --enable-benchmark
-
make
; make install
- manually provide link for binaries in system path at
/usr/local/bin/fsqc
to /home/hadaq/ltsm/install2025/bin/fsqc
; dito for fsqbench
- For proper compilation of DABC: we need to copy the
config.h
include files manually to
-
cp /home/hadaq/ltsm/config.h /home/hadaq/ltsm/install/include/ltsm/
-
cp /home/hadaq/fsq/config.h /home/hadaq/ltsm/install/include/fsq/
- This should be done by "make install", but it is not! TODO: feedback to Samuel.
Adjustment of tivoli config files:
-
/opt/tivoli/tsm/client/api/bin64/dsm.sys
- same as on other lxhadeb
-
/opt/tivoli/tsm/client/api/bin64/dsm.opt
- same as on other lxhadeb
DABC and stream frameworks
- Sources and all components under
/home/hadaq/soft/
-
/home/hadaq/soft/dabc/git
: source of dabc git clone of https://github.com/gsi-ee/dabc.git
-
/home/hadaq/soft/dabc/install/master
: most recent installation directory for dabc
-
/home/hadaq/soft/streamframe/git
: source of stream analysis framework, clone of https://github.com/gsi-ee/stream.git
-
/home/hadaq/soft/streamframe/install/master
: most recent installation directory of stream
- Proper rebuild is done by script
/home/hadaq/bin/update_dabc_install.sh
(link to repository version)
- The binaries, login scripts and xml configuration files are provided as soft links under
/home/hadaq/soft/dabc/bin
(directs to installation directory)
- The binaries and login scripts are finally linked into path at
/home/hadaq/bin
(direct to /home/hadaq/soft/dabc/bin
or into installation folders)
- The xml configuration files are finally linked into
/home/hadaq/oper/
(direct to /home/hadaq/soft/dabc/bin
)
--
Main.JoernAdamczewski - 2025-09-24 - 16:36
Scripts in hadaq path
hadaq@lxhadeb13:~/bin$ ls -l
total 24
lrwxrwxrwx 1 hadaq hades 76 Sep 24 17:30 archivecal_ltsm.sh -> /home/hadaq/soft/dabc/install/master/plugins/ltsm/app/archivecal_ltsm_fsq.sh
-rwxr-x--- 1 hadaq hades 7051 Sep 25 10:40 cleanup.pl
lrwxrwxrwx 1 hadaq hades 76 Sep 24 17:29 cleanup_evtbuild.pl -> /home/hadaq/soft/dabc/install/master/plugins/hadaq/hades/cleanup_evtbuild.pl
lrwxrwxrwx 1 hadaq hades 14 Sep 24 16:01 cmake -> /usr/bin/cmake
lrwxrwxrwx 1 hadaq hades 41 Sep 24 17:33 dabc_master -> /home/hadaq/soft/dabc/bin/dabc_exe.master
lrwxrwxrwx 1 hadaq hades 42 Sep 24 17:29 dabclogin -> /home/hadaq/soft/dabc/bin/dabclogin.master
lrwxrwxrwx 1 hadaq hades 76 Sep 24 17:31 restart_bnetmaster.sh -> /home/hadaq/soft/dabc/install/master/plugins/hadaq/app/restart_bnetmaster.sh
lrwxrwxrwx 1 hadaq hades 72 Sep 24 17:31 restart_rawmon.sh -> /home/hadaq/soft/dabc/install/master/plugins/hadaq/app/restart_rawmon.sh
lrwxrwxrwx 1 hadaq hades 55 Sep 24 17:30 streamlogin -> /home/hadaq/soft/streamframe/install/master/streamlogin
lrwxrwxrwx 1 hadaq hades 79 Sep 24 17:11 update_dabc_install.sh -> /home/hadaq/soft/dabc/install/master/plugins/hadaq/hades/update_dabc_install.sh
Note1: restart_bnetmaster.sh and restart_rawmon.sh are only required on dabc master machine. lxhadeb13 is meant as spare master for lxhadeb12
Note2: cleanup.pl is demon to remove oldest files from /data01 if necessary.
Config files in oper directory
hadaq@lxhadeb13:~/oper$ ls -l
total 0
lrwxrwxrwx 1 hadaq hades 56 Sep 24 17:37 BnetBuilderHadesFSD.master.xml -> /home/hadaq/soft/dabc/bin/BnetBuilderHadesFSD.master.xml
lrwxrwxrwx 1 hadaq hades 51 Sep 24 17:37 BnetInputHades.master.xml -> /home/hadaq/soft/dabc/bin/BnetInputHades.master.xml
Environment for raw monitor
This is on special eventbuilder servers lxhadeb12 (and spares lxhadeb12) only
The raw monitor is started by script
/home/hadaq/bin/ restart_rawmon.sh
. This used to be done remotely by a script from DAQ operator gui (
lxhadesdaq:/home/hadaq/trbsoft/daq/control/gui/
), but most recently the script was invoked directly by expert only
- working directory is
/data01/monitor/hades/dabc
- This contains
first.C
with analysis setup
- DABC runtime (with webserver) is provided by
hades.xml
- Logfile is written to
/home/hadaq/oper/DabcRawMon.log
crontab scripts
For user hadaq:
# m h dom mon dow command
@reboot /home/hadaq/bin/cleanup.pl -d &
0 0 * * * /home/hadaq/oper/move_oldruninfos.sh
*/4 * * * * /bin/ping -c1 -W1 192.168.104.250 >/dev/null 2>&1
The ping (to a non existing address) will frequently revive certain network switches. The
cleanup.pl is old demon to remove oldest hld files on
/data01
at high watermark.
For user root:
# m h dom mon dow command
*/2 * * * * /root/bin/set_eth_affinity_lxhadeb13.pl > /dev/null 2>&1
This script frequently resets the cpu affinity of the network cards interrupts to the desired assignments.
Otherwise these may be spoiled by the generic cpu management of the system and can get in conflict with the affinities set for the dabc eventbuilder processes (as defined in
/home/hadaq/trbsoft/hadesdaq/evtbuild/start_eb_gbe.pl
).
Copied from lxhadeb14 to lxhadeb13 (both are 64 cores now)->
adjusted for debian12 and different nic names!
network interface configurations
/etc/sysctl.d/hades_ip.conf
# JAM 15-06-2020 - tuning for event builders as tested by Michael T.
net.ipv4.ipfrag_high_thresh=41943040
net.ipv4.ipfrag_max_dist=1024
# JAM 8-2021 this was missing in defaults:
net.core.rmem_max=21299200
net.core.rmem_default=21299200
activate after change with
sysctl --system
nfs exports
- the
/home/hdaq/oper
directory needs to be exported to lxhadesdaq for optionally export runinfo files to Oracle. Also export it to hadesp33 for any hmon issues with TDC calibration files.
- the /data01 may be exported to online analysis servers hadesp63, hadesp66, hadesp67
Done in
/etc/exports
(copy from any other lxhadeb*)
hadaq@lxhadeb09:~/soft/streamframe/git$ cat /etc/exports
# /etc/exports: the access control list for filesystems which may be exported
# to NFS clients. See exports(5).
#
# Example for NFSv2 and NFSv3:
# /srv/homes hostname1(rw,sync,no_subtree_check) hostname2(ro,sync,no_subtree_check)
#
# Example for NFSv4:
# /srv/nfs4 gss/krb5i(rw,sync,fsid=0,crossmnt,no_subtree_check)
# /srv/nfs4/homes gss/krb5i(rw,sync,no_subtree_check)
#
#old monitoring machine is now also EBserver/data01 192.168.105.13(rw,sync,no_root_squash,subtree_check)
/data01 192.168.103.126(rw,sync,no_root_squash,subtree_check)
/data01 192.168.103.141(rw,sync,no_root_squash,subtree_check)
/data01 192.168.103.146(rw,sync,no_root_squash,subtree_check)
/home/hadaq/oper 192.168.100.50(rw,sync,no_root_squash,subtree_check)
/home/hadaq/oper 192.168.105.51(rw,sync,no_root_squash,subtree_check)
/home/hadaq/oper 192.168.105.52(rw,sync,no_root_squash,subtree_check)
/home/hadaq/oper 192.168.100.110(rw,sync,no_root_squash,subtree_check)
/home/hadaq/oper hadesp79(rw,sync,no_root_squash,subtree_check)
NOTE: on lxhadeb13 this was already exisiting!
Refresh the export via "exportfs -ra".
redirection of syslog
The local syslog should be redirected to lxhadesdaq, so commone eventbuilder messages show up in file
/home/hadeslog/messages
.
How to do it?
Exchange of BNET master machine
To replace BNET master (currently lxahdeb12) by e.g. lxhadeb13, a number of adjustments must be done:
- in
/home/hadaq/trbsoft/hadesdaq/evtbuild/start_eb_gbe.pl
: modify variable my $bnetmaster = "lxhadeb12";
(l.413). This is forwarded as environment variable BNETMASTER when the dabc bnet input and builder processes are started and used from the configuration .xml scripts.
- in
/home/hadaq/trbsoft/hadesdaq/control/gui/eb/Restart_BNET_control
: modify hostname in statement ssh hadaq@lxhadeb12
- several scripts at
/home/hadaq/trbsoft/hadesdaq/hmon/
- hmon_eb_rate.pl: adjust
my $masterurl = 'http://lxhadeb12p:8099/';
- hmon_spill.pl: adjust
my $masterurl = 'http://lxhadeb12:8099/';
- hmon_eberrbits_withreset.pl: adjust
my $masterurl = 'http://lxhadeb12:8099/';
- permanent/hmon_archiver.pl: exchange hostname in
my $filetype = get("http://lxhadeb12:8099/Master/...
- permanent/hmon_eb_missing_data.pl: adjust
my $masterurl = 'http://lxhadeb12:8099/';
- permanent/hmon_eb_run.pl: adjust
my $masterurl = 'http://lxhadeb12:8099/';
Exchange of rawmon machine
To replace the rawmon host machine (currently lxahdeb12), a number of adjustments must be done:
- several scripts at
/home/hadaq/trbsoft/hadesdaq/hmon/
- hmon_tdc_totratio.pl adjust hostname in
get ("http://lxhadeb12:8090/HADES/Run/...;
- no startup button for rawmon anymore, so adjustments of control/gui script not necessary
--
JoernAdamczewski - 2025-09-24