[Simh] VMS Cluster boot on an unique container
Boucher, François
boucher.francois at uqam.ca
Thu Oct 20 14:16:35 EDT 2011
Hi!
I never succeeded to operate a cluster of machines that attaches directly
to the same disk drives file images, when a first node writes to the system
disk, the other node was not aware of the changes and it ended up in a filesystem
corruption. Perhaps I did not found the correct way to create/configure some sort of
dual-port scsi devices?
I did succeed to make a 4 member cluster of VAXes running VMS 5.5-2,
the first VAX that boots (named EXTRA) boots off SYS0
then with MOP/DECNET I get the 3 other satellites to boot from
the system disk (SYS1, SYS2, SYS5) that EXTRA shares.
To get there, you need to set and define your node characteristics,
and enable service on line QNA-0
(all under MCR NCP...)
$ mcr ncp show node livre chara
Node Volatile Characteristics as of 20-OCT-2011 13:59:15
Remote node = 1.17 (LIVRE)
Hardware address = AA-00-04-00-11-04
Tertiary loader = SYS$SYSTEM:TERTIARY_VMB.EXE
Load Assist Agent = SYS$SHARE:NISCS_LAA.EXE
Load Assist Parameter = DISK$XPL0:<SYS1.>
those above characteristics give the possibility to boot from device XQA0
via the network. Those must be defined in your SYS0 node (EXTRA for me).
If you have a large number of disks ( more that 16), then you need to attach
disks on more than one simulator, and have the allocation_class parameter
given a different number on each sim, to permit a device name differenciation
i.e: $1$DUA0 on first node, $2$DUA0: on the other node, etc.
I have a dedicated ethernet interface per simulated vax node.
View of Cluster from system ID 1040 node: EXTRA 20-OCT-2011 14:04:14
lqqqqqqqqqqqqqqqqqqqwqqqqqqqqqk
x SYSTEMS x MEMBERS x
tqqqqqqqqwqqqqqqqqqqnqqqqqqqqqu
x NODE x SOFTWARE x STATUS x
tqqqqqqqqnqqqqqqqqqqnqqqqqqqqqu
x EXTRA x VMS V5.5 x MEMBER x
x REGIE x VMS V5.5 x MEMBER x
x AGENT x VMS V5.5 x MEMBER x
x LIVRE x VMS V5.5 x MEMBER x
mqqqqqqqqvqqqqqqqqqqvqqqqqqqqqj
This setup is stable, used in a limited (15 users) production level environment.
( It is used to replaced a 4 node VAX6000-640 CI cluster with 30 disks on HSJ40s)
$ sh dev d
Device Device Error Volume Free Trans Mnt
Name Status Count Label Blocks Count Cnt
$1$DNFS0: (EXTRA) Online 0
$1$DUA0: (EXTRA) Mounted 0 XPL0 5081589 265 4
$1$DUA1: (EXTRA) Mounted 0 XPL1 4773056 16 4
$1$DUA2: (EXTRA) Mounted 0 XPL2 1790901 1 4
$1$DUA3: (EXTRA) Mounted 0 XPL3 4152564 1 4
$1$DUB0: (EXTRA) Mounted 0 XPL4 1784088 1 4
$1$DUB1: (EXTRA) Mounted 0 XPL5 2971136 1 4
$1$DUB2: (EXTRA) Mounted 0 XPL6 5037920 1 4
$1$DUB3: (EXTRA) Mounted 0 XPL7 6682480 6 4
$1$DUC0: (EXTRA) Mounted 0 XPL8 5143296 1 4
$1$DUC1: (EXTRA) Mounted 0 XPL9 6193545 1 4
$1$DUC2: (EXTRA) Mounted 0 SIG0 4448640 1 4
$1$DUC3: (EXTRA) Mounted 0 SIG1 7621800 1 4
$1$DUD0: (EXTRA) Mounted 0 SIG2 6523400 1 4
$1$DUD1: (EXTRA) Mounted 0 SIG4 7662944 1 4
$1$DUD2: (EXTRA) Online 0
$1$DUD3: (EXTRA) Online 0
$1$DYA0: (EXTRA) Online 0
$1$DYA1: (EXTRA) Online 0
$3$DUA0: (REGIE) Online 0
$3$DUA1: (REGIE) Mounted 0 GST0 7803450 58 4
$3$DUA2: (REGIE) Mounted 0 GST1 4774653 1 4
$3$DUA3: (REGIE) Mounted 0 GST2 5030984 1 4
$3$DUB0: (REGIE) Mounted 0 GST3 4047615 38 4
$3$DUB1: (REGIE) Mounted 0 GST4 3188790 1 4
$3$DUB2: (REGIE) Mounted 0 GST5 4643512 1 4
$3$DUB3: (REGIE) Mounted 0 GST6 4273912 1 4
$3$DUC0: (REGIE) Mounted 0 GST7 5676296 1 4
$3$DUC1: (REGIE) Mounted 0 GST8 4991736 1 4
$3$DUC2: (REGIE) Mounted 0 GST9 5110560 1 4
$3$DUC3: (REGIE) Mounted 0 GST10 6117712 1 4
$3$DUD0: (REGIE) Mounted 0 GST11 6100728 1 4
$3$DUD1: (REGIE) Mounted 0 GST12 1357120 1 4
$3$DUD2: (REGIE) Mounted 0 GST14 5467352 1 4
$3$DUD3: (REGIE) Mounted 0 GST15 4480552 1 4
I hope this helps!
François Boucher
Ingénieur, SITel, AlphaTeam
boucher.francois_at_uqam.ca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20111020/2673118f/attachment-0002.html>
More information about the Simh
mailing list