Users

users@lists.einsteintoolkit.org

3 participants
3984 discussions

Release machine tests
by Ian Hinder 04 May '15

04 May '15

Hi, There is a table of test results generated automatically after runs of the tests on many machines for the release. The tests are initiated by Erik, and then the results are automatically uploaded to this system: https://build.barrywardell.net/view/EinsteinToolkitMulti/job/EinsteinToolki… This provides a slightly different view of the results than the "official" page. -- Ian Hinder http://members.aei.mpg.de/ianhin

1 0

Optimised CarpetInterp2 interpolation
by Ian Hinder 04 May '15

04 May '15

Hi, I just created a ticket describing an optimisation to CarpetInterp2 which should be beneficial to people using multipatch: https://trac.einsteintoolkit.org/ticket/1771 -- Ian Hinder http://members.aei.mpg.de/ianhin

1 0

Einstein Toolkit Meeting Reminder
by Frank Loeffler 03 May '15

03 May '15

Hi Please consider joining the weekly Einstein Toolkit phone call at 10 am US central time on Mondays. As usual, you can find instructions how to join on the following web site: http://einsteintoolkit.org/community/support/ In short: the number is (+1) 225-578-4942 and the conference id is 118682#. Most notably we will prepare for a new release, but any other issue is also welcome. Frank Löffler

1 0

meeting minutes for 2015-04-27
by Roland Haas 01 May '15

01 May '15

present: Frank, Jonah, Roland, Josh, Ian, Erik, Christine Corbett-Moran (Caltech), Elo Mailing list issues: * affected users were set to "bounced" status * Frank suspects this was because CCT changed firewall rules ET release: * recursive dependency between MPI and hwloc due to routines in the thorn itself (not in the distribution tarball) * accepted solution: move respective routines into SystemStatistics thorn * add -lnuma to all linux based option lists * Erik to try and see if his testing scripts still runs, will use current einsteintoolkit.th from fro development branch * local machines/linux/osx should be tested now as well * ssl certificate of jenkis build machine is invalid, Ian will look into this (domain is owned by Barry Wardell) Stampede: * no updates * need to decide on which compiler to use for release testing * Ian thinks that second maintenance should have brought in intel 15 * Caltech has an allocation, Christine will try and follow up locally to see if someone has already used it and could give intel 15 a try Yours, Roland

2 1

Testing production systems
by Erik Schnetter 01 May '15

01 May '15

Frank's announcement that he is preparing a new ET release, and that there is currently a feature freeze, may or may not have made it to the mailing list. FYI: The semi-automated ET tests currently include Blue Waters, Carver, Datura, Edison, Gordon, Hopper, Mike, Pandora, Philip, SHC, Shelob, Stampede, Trestles, Zwicky, as well as a few internal workstations. See < https://build.barrywardell.net/view/EinsteinToolkitMulti/job/EinsteinToolki… >. -erik -- Erik Schnetter <schnetter(a)cct.lsu.edu> http://www.perimeterinstitute.ca/personal/eschnetter/

3 3

Re: [Users] [Einstein Toolkit] #1751: [Pull request: CactusUtils/WatchDog] new thorn to automatically terminate jobs that hang
by David Radice 29 Apr '15

29 Apr '15

LGPL licensing (or even more permissive) is fine for me. > On Apr 29, 2015, at 8:37 AM, Einstein Toolkit <trac-noreply(a)einsteintoolkit.org> wrote: > > #1751: [Pull request: CactusUtils/WatchDog] new thorn to automatically terminate > jobs that hang > ------------------------------------+--------------------------------------- > Reporter: dradice@… | Owner: dradice@… > Type: enhancement | Status: reviewed_ok > Priority: optional | Milestone: > Component: EinsteinToolkit thorn | Version: development version > Resolution: | Keywords: > ------------------------------------+--------------------------------------- > > Comment (by rhaas): > > This thorn is reviewed ok. Unless there are objections I will add it to > CactusUtils (same repo as Nice and Formaline) after swapping the license > file for LGPL (and checking with David this is ok once more) today or > tomorrow. > > -- > Ticket URL: <https://trac.einsteintoolkit.org/ticket/1751#comment:12> > Einstein Toolkit <http://einsteintoolkit.org> > The Einstein Toolkit

1 0

Update website tutorial
by Ian Hinder 29 Apr '15

29 Apr '15

Hi, The "Tutorial for New Users" linked from einsteintoolkit.org points to the Queen Bee tutorial (https://docs.einsteintoolkit.org/et-docs/Tutorial_for_New_Users) on the wiki, which then states that it cannot be run at the moment: "NOTE: QueenBee 1 was decommissioned, and QueenBee 2 is not officially open yet. Stay tuned for updates for QueenBee 2." There is also the "Simplified Tutorial for New Users" (https://docs.einsteintoolkit.org/et-docs/Simplified_Tutorial_for_New_Users) but this is not linked from the main web page, so is not discoverable. This tutorial, apart from being "simplified", is also different in that it targets workstations (Debian, Ubuntu, Linux-Mint or Fedora) rather than Queen Bee. So the difference in name of the tutorial does not reflect the difference in content. What is the status of the queen bee tutorial? Is it functional, and if not, will it ever be? I think it would be good to have a single tutorial, which is simple, and which works anywhere. In cases where customisations for different systems are needed, this can be mentioned in the tutorial. This tutorial could then be the one linked from the ET website. Comments? -- Ian Hinder http://members.aei.mpg.de/ianhin

2 1

Re: [Users] [Einstein Toolkit] #1755: Problem with building on Ubuntu 14.04
by Maxim Barkov 28 Apr '15

28 Apr '15

Dear Rhaas, Please find in attachment make.HWLOC.defn. I used nightly version. Release version 2014_11 not compiled proper. I have only cactus_sim, cactus_2 do not exist. Cheers, Maxim On 20 March 2015 at 23:08, Einstein Toolkit < trac-noreply(a)einsteintoolkit.org> wrote: > #1755: Problem with building on Ubuntu 14.04 > > ------------------------------------+--------------------------------------- > Reporter: maxim.barkov@… | Owner: > Type: defect | Status: new > Priority: major | Milestone: ET_2014_11 > Component: Cactus | Version: development version > Resolution: | Keywords: MPI build > > ------------------------------------+--------------------------------------- > > Comment (by rhaas): > > The error you see is caused by the code no being linked against the NUMA > library. Since hwloc's compilation scripts changed a bit between the last > release (Herschel) and the current development version, the first we'd > need to know if which version of the ET you are trying to use: Herschel or > the development version? If you followed the tutorial and downloaded > > https://bitbucket.org/einsteintoolkit/manifest/raw/ET_2014_11/einsteintoolk… > then you have Herschel. > > Can you post the content of the file > > /home/bmv/utils/soft/ET/Cactus/configs/2/bindings/Configuration/Capabilities/make.HWLOC.defn > please? It will show if numa is listed in HWLOC_LIBS . Also the output of > dpkg --list would be helpful (as an attachement, please). > > I am very surprised that it actually ran since errors would seem to have > prevented Cactus from linking properly. The only way this could happen > would be if some utility did not build properly (in which case you may > choose to ignore it). > > Actually looking at you command used (the "2&>" part) you seem to have > created a new configuration called "2" and the one that you successfully > submitted was the one called "sim". Can you check if you have a > bin/cactus_sim but no bin/cactus_2 , please? > > -- > Ticket URL: <https://trac.einsteintoolkit.org/ticket/1755#comment:3> > Einstein Toolkit <http://einsteintoolkit.org> > The Einstein Toolkit >

3 2

At the last stage of gmake, undefined reference to 'pthread_atfork'
by Daren 28 Apr '15

28 Apr '15

Hello everyone, I am trying to install Cactus on my linux machine. The second gmake command should compile and generate an exe file of Cactus. However I got this error message: ------------------------------ In function `ompi_warn_fork': ompi_mpi_init.c:(.text+0xf4a): undefined reference to `pthread_atfork' collect2: ld 回傳 1 gmake[1]: *** [MYPATH/exe/cactus_new] Error 1 gmake: *** [new] Error 2 ------------------------------ I googled this and found someone got this error,too. And the solution suggested was to "add pthread library to the LIBS of the configuration". This, I tried adding each of the following line individually 1 . "LIBS=pthread" 2. "LIBS+=pthread" 3. "PTHREADS=yes" in my PATH/configs/new/config-info file but none of the three line works Is there any way to link my pthread to the library configuration? or is there any other problem I should solved?

2 1

Einstein Toolkit mailing lists were down
by Frank Loeffler 28 Apr '15

28 Apr '15

Hi, I am sorry to inform you that the Einstein Toolkit mailing lists were not working correctly for the period of one week (details below). Emails still made it to the lists, to the archives, but not out to the members of the lists again. As far as I can see no one was unsubscribed, but instead set to 'nomail'. As you can read this email now, this was fixed. Please let me know if you still have issues with any of the Einstein Toolkit lists. Old emails will not be re-sent to the email lists. In order to see what was sent to the users mailing list in April, please consult the archives: http://lists.einsteintoolkit.org/pipermail/users/2015-April/thread.html I again would like to apologize for the outage, Frank Loeffler Technical details (only read on if you _really_ want to know): About a week ago, the university firewall admins apparently blocked outgoing mail traffic (port 25) for the special group of machines the ET server is in - without notice to anyone, not even the department admins. This would have been noticed earlier, if not most machines in that group either relayed through an "official" mailserver anyway, or used IPv6 (which the university forgot to block). The ET machine was one of the unlucky machines that tried IPv4, and didn't relay: it ran into timeouts for every outgoing mail, eventually leading to bounces, which eventually lead to the 'nomail' status on the mailing lists. This usually produces a loud scream to the list admins, but that is sent via email... In addition I missed a direct email from one of the ET maintainers notifying us of the issue because I didn't have a good internet connection unexpectedly. Murphy's Law. After figuring this all out (we don't have a "direct line" to the university admins as you'd think, just the department admins), we decided to use the relay.

5 6

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

Users