Portservers and Boot PCs

1
Portservers and Boot PCs

• Colin Higgs
• Controls Pikett Training
• 3 May 2007

2
Portservers and Boot PCs

• Theory
• Naming Conventions
• Locations
• Auto Save and Restore
• Troubleshooting

3
Portserver Architecture
16
1
Serial connection to portserver
Network connection
The portserver host controls rmc access by
mapping the IOC ports to local devices
/dev/ttyXXNN
4
Schematic Elements
5

Portserver Hardware
16
1

Serial connections
17
32
32-port TS
Network connections
1
16
16-port TS
Power and Link LEDs
6
Beamlines on the PSI Network
Beamline BootPC / portserver host is a virtual
machine on a blade on the BladeCentre in WSLB
Machine network bootpc / portserver cluster
(excl. RF)
One blade on the Blade Centre
7
Beamline Portservers
Our Etherlite EL-16 portservers and host software
unfortunately only allow a 2-digit name
8
Machine Portservers
Our Etherlite EL-16 portservers and host software
unfortunately only allows a 2-digit name
9
Beamline Bootpcs and Portserver Hosts
BladeCentre in WSLB
Old Beamline Network
(1) To be migrated to WSLB during the June 2007
Shutdown
10
Machine BootPC Cluster / Portserver Host

• Two servers in rack WSLA.102.8 (server room)
• High reliability (RAIDed disks, redundant power
  supply)
• Synchronized file systems, /home/ioc, /ioc
• Login ONLY to slsbpc

slsbpc
11
Portserver Hosts

• Machine Network
• slsbpc (soon portserver-ac) for all machine
  crates except for
• portserver-rf034 and portserver-rf12 (RF plants)
• Beamline Network
• Each beamline BootPC hosts the portserver
  software for that beamline.

12
Auto Save and Restore

• On each BootPC under
• /home/ioc/archiver_data/autoSR/ltcrate_namegt/ltfiles
  gt
• Backups to the fileserver, slsfs01
• cron job runs every 8 hours, files kept for one
  week
• /exchange/share/autoSR/ltbootpcgt.tgz.ltdaygt.lthourgt
• e.g. /exchange/share/autoSR/x02da-bpc.tgz.Tue.11

13
Links on the Web

• Navigate to
• SLS Controls Homepage
• then select Web Tools
• then select SLS Hosts

BootPCs
Portservers
14
Portserver and BootPC Web views
15
Troubleshooting

• Problems with only one IOC? (the most common
  problem)
• Can you ping the IOC? Is it a network problem?
• Reboot the IOC if you can. Try with rmc f
  ltcrategt
• Logon to the portserver host and kill the
  process associated with the rogue IOC.
  (Actually, this is what rmc f ltcrategt should do
  !) Please report problems with this option!!
• Open a new rmc session to the crate and reboot
  it.

16
Troubleshooting (contd.)

• Problems with multiple IOCs connected to one
  portserver?
• Does the portserver have power? Can you ping the
  portserver?
• If the portserver is dead another must be
  configured (never happened so far....).
• Logon to the portserver host as root.
• Is the portserver process alive?
• ps auxw grep X1 for X02DA, internal name X1
  remember!
• You should see something like this if it is alive
• ./usr/bin/dgrp/daemon/dgrp X1 129.129.99.21 e
  never
• Or
• /usr/bin/dgrp/config/dgrp_gui and check that the
  portserver status is ready and not idle
• Start the portserver (as root) if it is idle.
• Problems with multiple portservers?
• Does the SLS have power?
• Follow the procedure outlined above. Here you
  will see a list of portservers. Restart those
  that dont have status ready.

17
Remember

• All portservers are REAL
• All portserver hosts are VIRTUAL (except for
  Machine RF)
• All bootpcs are VIRTUAL (except for Machine RF
  and the old beamline network)
• Rebooting a new beamline network bootpc /
  portserver host will only affect one beamline,
  (but this must be done by root remotely!) And
  remember, autoSR for the other crates will be
  affected during this time.
• Rebooting an old beamline network bootpc /
  portserver host affects other beamlines
• PC5163 is shared (X04SA, X05LA and FEMTO)
• PC5164 is shared (X06SA, X07MA and X07DA)
• PC5166 is shared (X10SA and X11MA) (will be
  removed in June 2007)
• Please DO NOT REBOOT slsbpc or portserver-ac on
  the Machine network