Message boards :
CMS Application :
New Version v47.40
Message board moderation
Author | Message |
---|---|
Send message Joined: 12 Sep 14 Posts: 1064 Credit: 325,950 RAC: 278 |
Upgrading vboxwrapper to 26197 on windows which enables support for Virtual Box 5.1.2. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Multi-core? |
Send message Joined: 8 Apr 15 Posts: 738 Credit: 11,558,539 RAC: 1,940 |
Multi-core? It appears to be multi I have it d/ling right now http://lhcathomedev.cern.ch/vLHCathome-dev/results.php?hostid=612 Mad Scientist For Life |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Thanks! I will give it a go in a bit. |
Send message Joined: 12 Sep 14 Posts: 1064 Credit: 325,950 RAC: 278 |
Yes but with the new project preferences you can set the number of CPUs = 1 to make it single core. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Tried to run a 4 core task. Failed twice--->VM Completion Message: No jobs were available to run. EDIT: tried 2core task-->seems to work. |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
Tried to run a 4 core task. See my message -> http://lhcathomedev.cern.ch/vLHCathome-dev/forum_thread.php?id=291&postid=4000#4000 So maybe no coincidence. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Maybe not. I assigned 5GB of memory--so alack of memory is no the cause. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
A 2 core task runs fine. A 3 core task only runs 2 jobs in slot 1 and slot 3. EDIT: One has to keep in mind, that when multiple jobs are uploading at the same time large amounts, it is going to be a problem, as we had in the past. |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
I tested twice 2 CMS-tasks with 2 processors in each VM. All 4 tasks errors after about 9 minutes with EXIT_NO_SUB_TASKS (CMS-jobs available) Starting 2 single core CMS-tasks works fine. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
You are using vbox 5.0.26? I am using 5.1.2. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
A 3 core task only runs 2 jobs in slot 1 and slot 3. It picked up a 3rd job 29min later than the other two. |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
You are using vbox 5.0.26? Yeah, I'll upgrade to 5.1.2 to test it with vboxwrapper version 26197 (only available for Windows yet). After the upgrade I'll start 1 dual core CMS without memory extention (no app_config), so 2048MB default RAM. |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
Yeah, I'll upgrade to 5.1.2 to test it with vboxwrapper version 26197 (only available for Windows yet). Done. 1 dual core CMS with default RAM doesn't get jobs either. I'll use a app_config now with RAM set to 3072 MB. http://lhcathomedev.cern.ch/vLHCathome-dev/result.php?resultid=235756 Condor StartLog: 08/09/16 13:54:48 ****************************************************** 08/09/16 13:54:48 ** condor_startd (CONDOR_STARTD) STARTING UP 08/09/16 13:54:48 ** /usr/sbin/condor_startd 08/09/16 13:54:48 ** SubsystemInfo: name=STARTD type=STARTD(7) class=DAEMON(1) 08/09/16 13:54:48 ** Configuration: subsystem:STARTD local:<NONE> class:DAEMON 08/09/16 13:54:48 ** $CondorVersion: 8.4.8 Jun 30 2016 BuildID: 373513 $ 08/09/16 13:54:48 ** $CondorPlatform: x86_64_RedHat6 $ 08/09/16 13:54:48 ** PID = 4156 08/09/16 13:54:48 ** Log last touched time unavailable (No such file or directory) 08/09/16 13:54:48 ****************************************************** 08/09/16 13:54:48 Using config source: /etc/condor/condor_config 08/09/16 13:54:48 Using local config sources: 08/09/16 13:54:48 /etc/condor/config.d/10_security.config 08/09/16 13:54:48 /etc/condor/config.d/14_network.config 08/09/16 13:54:48 /etc/condor/config.d/20_workernode.config 08/09/16 13:54:48 /etc/condor/config.d/30_lease.config 08/09/16 13:54:48 /etc/condor/config.d/35_cms.config 08/09/16 13:54:48 /etc/condor/config.d/40_ccb.config 08/09/16 13:54:48 /etc/condor/condor_config.local 08/09/16 13:54:48 config Macros = 153, Sorted = 153, StringBytes = 5980, TablesBytes = 5604 08/09/16 13:54:48 CLASSAD_CACHING is ENABLED 08/09/16 13:54:48 Daemon Log is logging: D_ALWAYS D_ERROR 08/09/16 13:54:48 Daemoncore: Listening at <10.0.2.15:29199> on TCP (ReliSock). 08/09/16 13:54:48 DaemonCore: command socket at <10.0.2.15:29199?addrs=10.0.2.15-29199&noUDP> 08/09/16 13:54:48 DaemonCore: private command socket at <10.0.2.15:29199?addrs=10.0.2.15-29199> 08/09/16 13:55:09 CCBListener: registered with CCB server lcggwms02.gridpp.rl.ac.uk:9623 as ccbid 130.246.180.120:9623#521719 08/09/16 13:55:10 HibernationSupportedStates invalid '' in ad from hibernation plugin /usr/libexec/condor/condor_power_state 08/09/16 13:55:10 VM-gahp server reported an internal error 08/09/16 13:55:10 VM universe will be tested to check if it is available 08/09/16 13:55:10 History file rotation is enabled. 08/09/16 13:55:10 Maximum history file size is: 20971520 bytes 08/09/16 13:55:10 Number of rotated history files is: 2 08/09/16 13:55:10 Allocating auto shares for slot type 0: Cpus: auto, Memory: auto, Swap: auto, Disk: auto slot type 0: Cpus: 1.000000, Memory: 1500, Swap: 50.00%, Disk: 50.00% slot type 0: Cpus: 1.000000, Memory: 1500, Swap: 50.00%, Disk: 50.00% 08/09/16 13:55:10 slot1: New machine resource allocated 08/09/16 13:55:10 Setting up slot pairings 08/09/16 13:55:10 slot2: New machine resource allocated 08/09/16 13:55:10 Setting up slot pairings 08/09/16 13:55:10 CronJobList: Adding job 'mips' 08/09/16 13:55:10 CronJobList: Adding job 'kflops' 08/09/16 13:55:10 CronJob: Initializing job 'mips' (/usr/libexec/condor/condor_mips) 08/09/16 13:55:10 CronJob: Initializing job 'kflops' (/usr/libexec/condor/condor_kflops) 08/09/16 13:55:10 slot1: State change: IS_OWNER is false 08/09/16 13:55:10 slot1: Changing state: Owner -> Unclaimed 08/09/16 13:55:10 State change: RunBenchmarks is TRUE 08/09/16 13:55:10 slot1: Changing activity: Idle -> Benchmarking 08/09/16 13:55:10 BenchMgr:StartBenchmarks() 08/09/16 13:55:10 slot2: State change: IS_OWNER is false 08/09/16 13:55:10 slot2: Changing state: Owner -> Unclaimed 08/09/16 13:55:10 State change: RunBenchmarks is TRUE 08/09/16 13:55:10 slot2: Changing activity: Idle -> Benchmarking 08/09/16 13:55:10 slot2: Changing activity: Benchmarking -> Idle 08/09/16 13:55:28 State change: benchmarks completed 08/09/16 13:55:28 slot1: Changing activity: Benchmarking -> Idle |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
I'll use a app_config now with RAM set to 3072 MB. Immediately starting with 2 cmsRun's. |
Send message Joined: 12 Sep 14 Posts: 1064 Credit: 325,950 RAC: 278 |
The memory is divided by the number of slots and there is probably a memory requirement in the job description use by the matchmaking. Memory scaling is required but I am still waiting to hear back on how to do this for a project with multiple applications. |
Send message Joined: 13 Feb 15 Posts: 1178 Credit: 810,985 RAC: 2,009 |
The memory is divided by the number of slots and there is probably a memory requirement in the job description... 2 cmsRun's in 1 VM running synchronous use almost 3GB of memory, but no Swap is used at all. |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
I got a 4 core VM started with 5632MB of memory. Fist it started 2 jobs and 20min later another two. Is that delay deliberate? |
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
I have a 4 core task running, where slot 2 does not start a new job,even though the 12h mark has not been reached. EDIT:Has the cutoff time been changed? I have now only 1 job running! |
©2024 CERN