The Infrastructure Unit operates the following general services across three sites: Informatics Forum, Appleton Tower, JCMB. Each site is set up so that it can operate as autonomously as possible, while at the same time providing redundant services to the other sites.
For the decant out of Appleton Tower we will be operating the Forrest Hill and Wilkie buildings as virtual floors of AT, to minimise the amount of setup and reinstallation work required. There will be edge switches only, with core infrastructure being provided from Appleton Tower. There are separate pages showning the proposed network arrangements for Forrest Hill and Wilkie.
Each site's ether switch configuration has been tailored to the particular circumstances. At the time of writing (September 2014) we have 164 network switches in the Forum, 43 switches in Appleton Tower, and 2 switches in JCMB. JCMB and the Forum also each have pair of FibreChannel fabrics. The Appleton Tower fabric has been discontinued, with the switches now being used as spares.
NOTE that 5600 FC switches are now unobtainable, and as a result of various failures we now have only one remaining FC "spare" switch. This "spare" 5600 also has fewer ports available than does the remaining KB 5800, which isn't currently a problem but does need to be kept in mind.
The Forum and Appleton Tower each have four network infrastructure machines, as follows:
At JCMB we have only three machines: one combining the first three roles above, a second acting as external nameserver, and a third acting primarily as console server but set up so as to be able to take on the network roles if required.
Each site has one "console server" machine, acting as a central point for all of the site's IPMI, KVM and Lantronix-serial consoles. In addition, they act as console server for each other; and in a few cases we have an off-site console server set up for critical machines. There's also a console server on a VM for the Forum self-managed server room consoles, primarily to make access control easier.
A nagios monitoring service is integrated with our machine configuration system. We have two nagios machines:
The primary is a real machine, to minimise dependencies, and the secondary a VM.
We operate a chat service, as this is also used by the nagios system to send alerts. The service runs on a VM.
This service will be transferred to the Services Unit in due course.
We use kerberos for authentication. There is one master KDC, in the Forum, and one slave KDC at each site. The iFriend master KDC is in the Forum, with a slave in Appleton Tower, and an additional slave at KB (this one is non-operational and purely to provide a backup away from the central area)
Although we don't make a lot of use of kx509 at the moment, we still run a kx509 service. There are currently two KCAs, both on VMs.
Most web authentication now uses cosign. We currently have cosign servers (physical machines) in the Forum and Appleton Tower. These also co-host the iFriend KDCs. These services are not suitable for co-locating with the main KDCs for security reasons.
Hosts and services requiring a locally-signed X.509 certificate obtain this using the sixkts service. As this is not a high-availability requirement, we currently have one sixkts server on a VM.
The OpenLDAP master is currently in the Forum, and there are site slaves in the Forum, Appleton Tower and JCMB. In addition, all DICE machines also currently run a full slave configuration. Note that this is currently under review, and arrangements may change as a result.
The prometheus system runs on a VM in the Forum. This would not be an easy service to replicate, but immediate availability is not a requirement and we could move the service to another machine should the main one be unavailable. This could be either the development server or the JCMB OpenLDAP server, or indeed another VM, as decided at the time.
|Name||Type||Location||Role||S/N||P/O & date||Warranty||Replace||UPS|
|(GPS receiver)||ACUTIME 2000||Forum roof||Timestamps for NTP S1||82175548||a628299 2005-01-31||??||as and when||(Included here for completeness)|
|crystal||GX745||Forum 5A closet||NTP S1, DHCP (B)||428903J||ikb0153 2007-06-22||3Y||as and when||"Real" serial port required|
|blackwell||R610||AT server room||LDAP site-slave||9GD2D4J||inf0488 2009-06-30||3Y||2014-15 |
|darwin||R200||AT server room||extDNS, extNTP||H660D4J||inf0488 2009-06-30||3Y||??||Formerly "ancerl"|
|fenrir||R200||Forum server room||KDC, AFSDB||J660D4J||inf0488 2009-06-30||3Y||2014-15|
|linnaeus||R200||Forum server room||extDNS, extNTP||C560D4J||inf0488 2009-06-30||3Y||??|| ||Formerly "hickox"|
|mckinley||R610||Forum server room||LDAP site-slave||BGD2D4J||inf0488 2009-06-30||3Y||2014-15 |
|hati||R210||JCMB server room||KDC, AFSDB||GJ6TJ4J||inf0622 2009-11-02||3Y||2014-15|
|skoll||R210||AT server room||KDC, AFSDB||FJ6TJ4J||inf0622 2009-11-02||3Y||2014-15|
|otaka||DL180||AT server room||AT netInf, DHCP||CZ30291JCP||inf0953 2010-07-09||4Y||2014-15||1kVA AS0614310419||Replace norrington with 10Gbps DA and roll down?|
|reeves||DL180||Forum server room||LDAP master||CZ30301P1K||inf0953 2010-07-09||4Y||2014-15? |
|WARRANTY EXPIRED ABOVE HERE|
|abbado||DL180||Forum server room||Forum netInf, DHCP||CZ3115CJ62||inf1175 2011-03-31||5Y||2016-17||3kVA JS0511022795|
3kVA XL QS0348111013
|cockerel||DL180||Forum server room||Nagios master||CZ3121H23L||uoe26808 2011-05-18||5Y||2016-17|
|hall||DL180||JCMB server room||LDAP site-slave/prometheus DR||CZ3121H23F||uoe26808 2011-05-18||5Y||2016-17 |
|tycho||DL180||Forum server room||loghost||CZ3121H23H||uoe26808 2011-05-18||5Y||2016-17||750VA AS0444223639|
|slatkin||R310||JCMB server room||KB netServ, consoles, ifriend KDC slave (unused)||4WL3C5J||inf1642 2012-05-10||5Y||2017-18||3kVA JS0510018437|
|elder||R320||JCMB server room||KB extRt, netInf, DHCP||BB3YC5J||inf1748 2012-06-18||5Y||2017-18||3kVA JS0511022966|
|norrington||R320||AT server room||AT extRt||CB3YC5J||inf1748 2012-06-18||5Y||2017-18||1400VA XL QS0322110541
1400VA XL QS0322210008
|Roll down to netInf and add new 10Gbps DA machine as extRt?|
|bevan||R320||Forum server room||KDC master||B15CD5J||inf1771 2012-06-22||5Y||2017-18 |
|blatiere||R320||Forum server room||Forum consoles master, DHCP||J9MM9X1||inf2541 2013-03-28||5Y||2018-19|
|hanlon||R210||Forum server room||cosign, iFriend KDC master||6LQM9X1||inf2539 2013-03-28||5Y||2018-19|
|mcintyre||R210||AT server room||cosign, iFriend KDC slave||31JFWX1||inf2539 2013-03-28||5Y||2018-19|
|rattle||R320||Forum server room||Forum netServ, site DNS, OpenVPN, DHCP (static)||G1KL9X1||inf2540 2013-03-28||5Y||2018-19||3kVA XL QS0348111013|
|gatti||R320||AT server room||AT netServ, site DNS, OpenVPN, DHCP (static)||BKZKT02||inf3726 2014-03-18||5Y||2019-20||1400VA XL QS0322110541
1400VA XL QS0322210008
|grepon||R320||AT server room||AT consoles||4KZKT02||inf3726 2014-03-18||5Y||2019-20|
|knussen||R320||Forum server room||Forum extRt||9CKKT02||inf3719 2014-03-18||5Y||2019-2||??|
|descartes||VM||KB||Test RADIUS server|
|euclid||VM||AT||Test RADIUS server|
|peigne||VM||IF||Forum self-managed consoles|
|pythagoras||VM||IF||Test RADIUS server|
Machines in the Forum and AT server rooms are covered by the inbuilt UPSes, and are shown with a blank in the column unless they have some additional provision. Machines in the JCMB server room may be powered by one of the "rack" UPSes, and in this case are shown with a blank in the column unless they have some additional provision.
|Name||Type||Location||Role||S/N||P/O & date||Comments|
|fc00||SANbox 5600||IF-B.02||0835C00819||ikb0626 2008-09-05|
|fc01||SANbox 5600||IF-B.02||0834C00021||ikb0626 2008-09-05|
|fc0a||SANbox 5600||AT server room||0834C00108||ikb0626 2008-09-05||Taken out of service as a possible spare|
|fc10||SANbox 5600||IF-B.02||0834C00018||ikb0626 2008-09-05|
|fc11||SANbox 5600||IF-B.02||0834C00113||ikb0626 2008-09-05|
|fc1a||SANbox 5600||AT server room||0834C00026||ikb0626 2008-09-05||Taken out of service and in use as a spare at KB|
|fc0||SANbox 5800||JCMB||1005F00659||inf0778 2010-03-09||PSU fault|
|fc1||SANbox 5800||JCMB||1005F00525||inf0778 2010-03-09|
NOTE that 5600 switches are now unobtainable, and as a result of various failures we now have only one remaining FC "spare" switch.
|Name||Type||Role||Location||S/N||P/O & date||Rating||Battery||Comments|
|SMART-UPS 1400 RM XL||AT server room||comms cabinet||QS0322110541||a602939 2003-06-06||1400VA||2011-06|
|SMART-UPS 1400 RM XL||AT server room||comms cabinet||QS0322210008||a602939 2003-06-06||1400VA||2011-06|
|Smart-UPS 1000 RM||AT server room||rack 2||AS0614310419||a637388 2006-06-27?||1kVA|
|Smart-UPS 3000 RM||AT3||in store||JS0510018446||a631778 2005-07-11||3kVA||2011-07|
|Smart-UPS 3000 RM||AT4||in store||JS0617023554||a637388 2006-06-27||3kVA||2014-10|
|Smart-UPS 3000 RM||AT5||in store||JS0510018447||a631778 2005-07-11||3kVA||2013-02||Battery moved from AT8 (RT#63576)|
|Smart-UPS 3000 RM||AT6||in store||JS0511022967||a631778 2005-07-11||3kVA||2014-10|
|Smart-UPS 3000 RM||AT7||in store||JS0714011678||ikb0141 2007-06-19||3kVA|
|Smart-UPS 3000 RM||AT8||in store||JS0714011688||ikb0141 2007-06-19||3kVA||2013-09|
|SMART-UPS 3000 RM XL||Forum core0 ("core")||comms racks||QS0348111013||a614789 2004-02-04||3kVA||2013-01|
|Smart-UPS 3000 RM||Forum core1 ("netInf")||comms racks||JS0511022795||a631778 2005-07-11||3kVA||2012-05|
|Smart-UPS 3000 RM||Forum core2 ("netServ")||comms racks||JS0617023553||a637388 2006-06-27||3kVA||2011-07?|
|Smart-UPS 750 RM||Forum loghost||comms racks||AS0444223639||a627391 2004-12-02||750VA|
|Smart-UPS 1500||Test & development||IF-2,36||YS0315121217||a609907 2003-08-14||1500VA||2013-03|
|Smart-UPS 3000 RM||JCMB server room||Rack 0||JS0617023553||??||3kVA|
|Smart-UPS 3000 RM||JCMB server room||Rack 1||JS0511022966||a631778 2005-07-11||3kVA|
|Smart-UPS 3000 RM||JCMB server room||Rack 2||JS0510018437||a631778 2005-07-11||3kVA|
|SMART-UPS 5000 RM DL4||JCMB server room||Rack 3||CS0543110262||??||5kVA|
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh