NodeTests

Tasks to do:

Initial work

(HP) Manual work

• Blank disk of node, wipe by:

 dd if=/dev/zero of=/dev/sda; sync

• Put MAC address into DHCP table on server
• PXE boot

Automatic from here (check list):

• Flash BIOS (Dos ) -> DONE
• Flash Firmware (Dos ) -> DONE
• Setup BIOS/IPMI ( (Dos )/Linux) -> DONE

• Clone/Install operating system ((FAI )) -> in progress
• Reboot -> DONE

Slave tests

Sensors

(HP) remotely get

• temperatures? -> DONE
• fan speeds? -> DONE
• voltages? -> DONE
• SMART values?

by using IPMI

Warnings

(all) Are there any warnings in

• dmesg
• /var/log/{messages|syslog}

Partitions

• correct partition tables, inode number customization -> DONE (config in FAI)

Clean?

• are /boot, /lib/modules clean? -> DONE

X/tools

• does startx work? Which WindowManagers? dwm -> DONE
• not standard way to boot -> DONE
• gcc/ddd/gdb/prof/gprof/valgrind/g++ + other vital tools on the head nodes??? -> DONE

Network

• networking runs wire-speed, full-duplex (netperf/netpipe)
• (VLAN?) wait for the core swicth
• correct identity for machine (hostname, IP) -> DONE

Power

(MS)

• shutdown -hf now (maybe only shutdown -h) -> DONE
• After shutdown, power cycle (disconnect, reconnect cable) box needs to stay off -> DONE
• shutdown -rf now reboots -> DONE
• unplug box, plug back in, box should stay off -> DONE
• remotely power on machines (IPMI, etherwake)? Under any of given conditions (except reboot) -> DONE
• cut UPS power for less than 60s, nodes should stay on -> DONE
• cut UPS pwer for more than 60s, nodes should shut down -> DONE
• What about not full UPS?

Time

• no files with dates in the future -> DONE

 cd /; touch temp.dat; find / -xdev -cnewer temp.dat

• no files with dates with "early dates" (e.g. 1980) -> DONE

  find / -xdev -type f -printf "%TY %p\n"| grep "^19[0-7][0-9] "

• ntp working? -> DONE

 ntpq -p

Benchmarks/Tests

• benchmarks run at full speed?
• disk speeds (guesstimate > 50 MB/s and > 800 MB/s) -> DONE

  hdparm -tT /dev/sda

• big file support (>2 GB) -> DONE
• /proc/meminfo should show full memory

Automount

• Automounting works? E.g. /net/s1234/data
• cd to automounted directory?

 cd /net/s1234/data

• Copy to/from automounted partitions (permissions for users/root correct - what is correct?)
• low NFS time-out values

rsh/ssh

• Can root on master rsh to any node? -> DONE
• Can ordinary rsh to any node? -> DONE
• Node to node rsh should work as well (host.allow/host.deny) -> DONE
• rsh uptime on any nodes? -> DONE

 master $ rsh s1234 uptime

Misc

• does /root/cloned-date (svn-tagnumber) exist? Maybe better in /etc? -> Puts date in /etc/cloned-date '-> DONE '
• does recloning preserve data? -> DONE '' if it is covered by fai softupdate
• does email work on the nodes (outgoing)? (X) -> DONE * garbage bag test. Computer starts to overheat and should shutdown cleanly. -> DONE