| |
| META TOPICPARENT | name="CemITSystem" |
Contents
Administration logs
Jan 2011
- Set up and installed new computer, cem15
- CentOS? 5.5, RAID5, 5.2 TB disk
- All standard EM software installed
- cem15data drive is nfs shared and exported
- users added on ad hoc basis
April 2011
- April 7 2011
- back up svn server from cem100 and move to cem01
- cem01 now running svn server (current revision=48) and apache server
- made very basic web page for cem01 which links to svn repos as well as sesame
- April 8 2011 -- rename computers as follows
- cem15 --> orpheus
- cem01 -->cronus
- cem02 --> priam
- cem03 --> agamemnon
- cem11 --> tlaloc
- cem12 --> minotaur
- made new hosts file listing these changes, and uploaded to svn server
- TO DO
- rename cem100 to athena -- this will be done after the 2dx meeting at nysbc, so thigns will not break
- DONE 04-27-11
- April 10 2011
- copy all of Kens' data from Tecnai to external hard drive
- delete original copy of tomograms from Tecnai hard drive, leaving only navigator maps
- April 11 2011 -- rename drives as follows:
- cem02bigdata --> helen
- cem02data --> hector
- cem02data2 --> hecuba
- cem03bigdata --> achilles
- cem03data --> menelaus
- cem03data2 --> iphigenia
- cem11data --> tonatiu
- cem12data --> theseus
- cem15data --> eurydice
- exported and mounted all newly named drives
- updated hosts file on jeol1230 pc to reflect these changes
- change backup scripts to reflect these changes
- found problem with leginon not talking to jerol1230 from minotaur (cem12) after change
- fixed by replacing minotaur entry with minotaur.nysbc.org in cd:\windows\system32\drivers\etc\hosts file
- verified that leginon is still working directly on jeol1230 and remotely from minotaur
- links made to old drive names
- cronus (cem01) now viewable from outside as cronus.nysbc.org
- verified by browsing cronus.nysbc.org from public network
- TO DO
- rename dl380 to arachne, and make link
- finish renaming, after backup server installed and running
- rename cem02bigdata_backup to paris
- rename cem03bigdata_backup to odysseus
- find use for external USB RAID on minotaur, since it will no longer be used for backup. Kens data???
- April 20 2011
- updated IMOD to IMOD 4.1.9 (current version) on /cryoem
- /cryoem/IMOD now refers to this version
- /cryoem/IMOD3 referes to imod 3.13
- updated .cshrc fil on orpheus, agamemnon, minotaur to reflect this
- note that priam (centos4) will likely not run this version
- April 21 2011
- Checked Tecnai F20 computer (Win 2000)
- 19 GB used out of 34 GB
- deleted ~3 GB of images from old tests
- deleted empty folders for several users who are no longer here
- deleted serialem links for users who are no longer here
- shortcut to feispy and teispy found in trash; restored to desktop
- Current disk usage =~ 16 GB, just under 50% capacity
- TODO -- install internal drive temporarily for backup. Do on maintenance day
- April 27 2011
- updated hostname of cem100 to athena
- remounted cem100/data as athena/data on cronus,priam,agamemnon,minotaur
- ensured that hosts file is updated
- updated hosts file on jeol1230 computer -- cem100 is now athena
- updated sinedon.cfg on j1230 and minotaur -- now has athena as db host
- todo -- fix mounting of cem100 as athena on j1230 -- ckim is using
May 2011
- May 24 2011
- updated imod on /cryoem to IMOD 4.1.10
- installed SerialEM 3 on Tecnai F20
- May 26 2011
- testing of 32-bit helix programs in kubuntu 9 (lab laptop) to see which need recompiling: listing of progs which do not work
- divctf-working-2006 (ELF class 64 shared library)
- hlxavg (ELF class 64 shared library)
- hlxresidual.pyc (should not be included -- need hlxresidual.py instead)
- maskparam (ELF class 64 shared library)
- plotphase (ELF class 64 shared library)
July 2011
- update priam to CENTOS5
- repalced root drive with 500GB drive
- root partition 100GB
- hector partition 350 GB
- memory upgraded to 12GB
- new GPU installed from cem11 (Ruben's) so GPU can be used for calculations (CUDA)
- Zeiss Windows 2000PC -- removed Norton antivirus
- disk nearly 100% full -- this saved 1 GB of space
- computer is isolated from web, so viruses should not be a problem
August 2011
- Aug 17-19 --
- update temimps2twiki.py on athena -- now downloads all attachments (ppt,pdf,zip,...)
- now uses BeautifulSoup? to do most searching
- need to update twiki script to do the same
- Aug 24 -- set up backup server tartarus:
- running CENTOS 6
- partitioned backup partiton 23.9 TB (RAID5) with GPT (parted)
- "mkpart ext4 1 -1" --> makes a single partiton using all of disk space, named it ext4
- xfs format (ext4 currently has 16TB maximum size)
- mount as /pithos/
- installed rdiff-backup:
- needed to enable rpmforge repository
- installed yum-priorities -- put CentOS? priorities at 1, rpmforge at 10 to prevent rpmforge from overtaking centos
- no repository for rdiff backup: installation required manually installing
- pylibacl-0.5.0-1.el6.x86_64.rpm
- pyxattr-0.5.0-1.el6.x86_64.rpm
- gcc using yum
- python-devel using yum
- librsync-devel using yum
- then needed to manually install and compile rdiff-backup-1.2.8 (rpm did not work)
- installed in /usr/local/
- Aug 23
- installed gromus on agamemnon
- required installation of fftw3 libraries in /usr/local
December 2011
- Agamemnon: menelaus drive moved from /dev/sda5 to /cem03_bigdata_backup
- menelaus now has lots of space, and root partition can be expanded to take over all of /dev/sda
- Nicolas, Mariena,Ralph PC's are backed up to /hector
January 2012
- Jan 3
- New PC set up: rhea (one of the old Win2000 Dell PC's)
- purpose is to
- act as a spare workstation
- drive added - hades, which is exported onto windows drives using samba.
- EM users who want to copy onto USB flash disks the same day can do safely so via this PC
- Jan 16
- Priam: mounted former /cemo2_bigdatabackup as /paris (1.4 TB drive)
- /paris exported and mounted on eurydice, agamemnon, minotaur,cem11
- /paris set up to be backed up daily on tartarus
- Agamemnon : root partiton resized to 230 GB, from 30
- root was very full (2Gb free), so expanded ino space formerly used by menelaus
March 2012
- March 14
- Set up Ed's PC - Centos5
- install php mrc tools:
- copied source from cronus (/usr/include/php/ext/mrc) and unzipped in identical location
- downloaded FFTW2 library source (NOT FFTW3 as is mention in docs) and installed:
./configure --enable-float --enable-type-prefix --enable-shared
make
make install
-
-
- libs were put in /usr/local/lib, however, they are 64-bit libraries so made links in /usr/local/lib64/
- compiled without error however fft power spectrum was invisible
November 2012
Setting up new Hp 620z workstation
- system
- two six-core Xeon processors (12 cores total)
- 32 GB RAM (4* 8 GB; system came with 6*2 GB but this is incompatible; there are 8 slots still free for a max of 96 GB RAM)
- 500 GB HD with Windows 7-64 pro installed; 3 extra disks Western Digital 2GB Black 7200rpm
- NVidia Quadro 2000 GPU (192 CUDA cores)
- hardware setup
- Installed RAM and GPU; boots windows
- only 2 bays left for the 3 drives; need to convert one 5.25" bay to 3.5"; adapter on order (Nov 20)
- Attempted to install all RAM simultaneously -- no boot. Removing origina RAM allowed boot.
- linux install
- Choosing CentOS? 6.3 as a modern system
- system could not boot from DVD: has EFI instead of old-style BIOS
- therefore had to choose install from Centos6.3-64 EFI minimal CD
linux install details
DEVICE="eth0"
HWADDR=specifc mac address
NM_CONTROLLED="no"
ONBOOT="yes"
BOOTPROTO="static"
IPADDR=ip address here.
NETMASK=netmask here.
-
- /etc/sysconfig/network and /etc/resolv.conf left unchanged (resol.conf was a 0 length file)
- restart network
/etc/init.d/network restart
- install some packages
- yum install kdebase.x86_64 -- get some graphics at least
- yum install firefox -- get a browser
- yum install xorg-x11-server-xorg.X86_64 - - get startx to work
- yum install xorg-x11-xinit-session-1.09-13.el6.x86-64 --
- sudo nano vim wget system-config-network-tui kernel-headers make gcc ntp tcsh emacs kernel-devel
- install Nvidia driver NVIDIA-Linux-x86_64-310.19.run
- problem: complains about Nouveau kernel driver. Alllowed it to disable this.
- to undo this, need to delete /etd/modprobe.d/nvidia-installer-disable-nouveu.conf
- reboot and try again
- still enabled!
- yum update -y
- edit /boot/efi/EFI/redhat/grub.conf anmd append rdblacklist=nouveau to kernal boot command to disable
- reboot. /sbin/lsmod | grep nouveau indicates that it was not loaded
- install NVIDIA module and run startx. success!
- edit /sbin/inittab to make level 5 the default boot
- konsole terminal seems messed up a bit
- yum groupinstall "Graphical Administration Tools" -- gui to admin tasks
- now can use gpk-application to install required software
desktop details
- switchdesk, used to switch between gnome, kde, and others is no longer available
- attempts to make this available by making /etc/sysconfig/desktop did not work
- in home directory, make the file ~/.dmrc as follows:
[Desktop]
Session=kde
Language=en_US.utf8
- needed to install 32-bit libraries to get web to work:
- libXmu-1.0.5-1.el6.i686
- libXt-1.0.7-1.el6.i686
- libXtst-1.0.99.2-3.el6.i686
- libXp-1.0.0-15.1.el6.i686
November 26 2012
- yum update -y on orpheus, installed latest Nvidia driver, and rebooted
January 2 2013
Setup Styx /computer for storage
- Put together Intel Core i3 CPU + motherboard + 8 GB RAM + 450W power supply
- Motherboard: ASUS P8H61-MLE/CSM
- RAM: 2x 4GB
- Install one boot drive: 500 GB from Kakoli's old RAID system
- Install 5X 2TB drives for storage (configure as RAID5; see below)
- No CD/DVD/floppy
details of system install
- Install CENTOS 6.3 Minimal from external USB DVD drive
- network configured manually as described above
- yum update -y
- reboot
- install other essential packages as detailed above
preparation of hades RAID5 disk
- hades: RAID 5 using 5x 2 TB Western Digital drives
- instructions on making these drives a RAID (hdb,hdc,hdd,hde,hdf)
> sudo mdadm --create /dev/md0 --level=5 --raid-devices=5 /dev/sd[bcdef]1
> watch cat /proc/mdstat (wait until done, about 450 minutes)
DEVICE /dev/sdf1 /dev/sde1 /dev/sdd1 /dev/sdc1 /dev/sdb1
ARRAY /dev/md0 level=raid5 num-devices=5 metadata=1.2 name=styx.nysbc.org:0 UUID=2d2b02fa:a1970439:77152d3a:7d64280e
devices=/dev/sdb1,/dev/sdc1,/dev/sdd1,/dev/sde1,/dev/sdf1
MAILADDR rice@nysbc.org
> sudo chkconfig mdmonitor on
- make a logical volume, format it and mount it
> sudo pvcreate /dev/md0 -- creates a physical volume
> sudo vgvreate HADES /dev/md0 -- creates a volume group on that physical volume
> sudo lvcreate -L 7.27T -n hades HADES -- creates a logical volume in that volume group
> sudo mkfs.ext4 /dev/HADES/hades -- makes the filesystem on the logical volume
> sudo mkdir /hades
> sudo mount /dev/HADES/hades /hades
- nfs configured to export /hades
Setup new workstation fir rice Dec 2014
* Configuration:
-
- AMD FX-6350 six core processor
- 16 GB RAM
- 650 W power supply
- ATI graphics card
- Crucial MX100 512 GB SSD for OS
- 4X 3 TB WD Red drives for storage
- originally bought as dataserver; found performacs = 2X that of current CEM11 (Fabiana's old workstation)
- Will repurpose cem11 as data server (with Nvidia card) and use new system as my desktop
- keep paris,cem11data in cem11 computer, keep runnina as headless server
- to do later: upgrade to CentOS? 6
- Standard Linux Mint 17.1 install with cinnamon
- set up 3 TB drives as RAID 1 0
> sudo parted /dev/sdb
> mklabel gpt
> mkpart primary ext4 0% 100%
> quit
- repat for /dev/sdc,/dev/sdd/, /dev/sde
- Install Linux Mint 17.1 onto 512 GB SDD
- Netowrking not working -- ifcofig seems ok but ping does not work
- after MUCH strugging, solution was
- go to BIOS
- enable IOMMU
- restart
- edit /etc/default grub: change empty quotes in line to:
GRUB_CMDLINE_LINUX="iommu=soft"
sudo update-grub
- Now networking, USB2 and USB3 all work
- install new drive : 2 TB spare drive
- this renumbered disks:
- boot is /dev/sdc
- spare is /dev/sdb
- make RAID on sda,sdd,sde,sdf
- re-format sda1,sdd1,sde1,sdf1 as ext4
- sudp apt-get install mdadm
sudo mdadm -v --create /dev/md0 --level=raid10 --raid-devices=4 /dev/sda1 /dev/sdd1 /dev/sde1 /dev/sdf1
mdadm: layout defaults to n2
mdadm: layout defaults to n2
mdadm: chunk size defaults to 512K
mdadm: /dev/sda1 appears to contain an ext2fs file system
size=-1364702208K mtime=Wed Dec 31 19:00:00 1969
mdadm: /dev/sdd1 appears to contain an ext2fs file system
size=-1364702208K mtime=Wed Dec 31 19:00:00 1969
mdadm: /dev/sde1 appears to contain an ext2fs file system
size=-1364702208K mtime=Wed Dec 31 19:00:00 1969
mdadm: /dev/sdf1 appears to contain an ext2fs file system
size=-1364702208K mtime=Wed Dec 31 19:00:00 1969
mdadm: size set to 2930133504K
Continue creating array? y
mdadm: Defaulting to version 1.2 metadata
mdadm: array /dev/md0 started.
sudo pvcreate /dev/md0
sudo vgcreate PARKING /dev/md0
sudo lvcreate -l100%FREE -n parking PARKING
sudo mkfs.ext4 /dev/PARKING/parking
sudo mkdir /parking
sudo mount /dev/PARKING/parking /parking
- mounted OK
- for debian-type, need to manually edit mdadm.conf as follows (as root)
mdadm --examine --scan --config=mdadm.conf >> /etc/mdadm/mdadm.conf
- NOTE
- tried flag "amd_iommu=on" : USB2 and ethernet still work but no USB3
- left as iommu=soft
|
| |
[rice@styx ~]$ sudo hdparm -i /dev/sda | grep Serial
Model=WDC, FwRev=02.03B03, SerialNo=WD-WCASY7050076
[rice@styx ~]$ sudo hdparm -i /dev/sdb | grep Serial
Model=WDC, FwRev=05.01D05, SerialNo=WD-WMAWP0561880
[rice@styx ~]$ sudo hdparm -i /dev/sdc | grep Serial
Model=WDC, FwRev=80.00A80, SerialNo=WD-WMC1T0928093
[rice@styx ~]$ sudo hdparm -i /dev/sdd | grep Serial
Model=WDC, FwRev=80.00A80, SerialNo=WD-WMC1T1328932
[rice@styx ~]$ sudo hdparm -i /dev/sde | grep Serial
Model=WDC, FwRev=80.00A80, SerialNo=WD-WMC1T0906133 |