MOVING unicorn TO SETER

PRELIMINARY NOTE: The first load (April from Osiris to Seter will be onto the SCSI disk that comes with Seter. On April 24 The Unix team will erase that info and ATTACH the SSA disk from Osiris to Seter: therefore no ‘load’ as such.

Schedule meeting with Roger, Barrie, Nella, Janice (LP)
Notes to Angus, Joseph, Kin, Don Koz, Kirkham and Project Team (LP)
Rename the B80 to seter (Kin)
Get seter upgraded - add a CPU card so it’s dual processor, order and add SSA card (Joseph/IBM/Kin)
Set up filesystems to match osiris (Kin)
Clone osiris to seter (Kin/Joseph)
Osiris is 136.159.86.205
Seter is 136.159.86.206

TEST PHASE

Test everything (LP and Teamheads)

Message to allir-m re scheduling and plans (LP)

Message on the Library webpages regarding no user services for half a day? (Eric Tull)

Schedule extra full system backup on osiris for just after midnight the morning of Wednesday April 24 (LP ask Angus) (Talk to Joseph about this -- disk to disk as well I assume)

Reminder message about when to stop cataloguing, acquisitions, circulation, etc. (LP)

Tuesday APRIL 23

11:45 p.m. via a cronjob, replace the schedlist with some suspended reports <adutext, rebuilduath, reorgtext, dtb, cash> (Nella)

4:30 or later: Disable user services on osiris AND neter in both gateways. Do halt-and-run on neter. No more cataloguing or other transactions, but still in use for public access.

Comment out crontab lines for ID and FRS interfaces (Roger/Barrie)

5:00 Stop incoming mail files (send to file) (Roger/Barrie)

Disable PIN check on the Tuesday night, April 23 (Barrie)

Comment out hist_to_horus after 9. Ditto hogwatch, verify, etc.

Do a last-minute transaction so when system comes up I can see if it's there

Authentication:libauth has to point to the test machine on the campus DNS. Next time remember to ask Network Services to set the time to die to a few minutes the day preceding this maneuver.

WED. APRIL 24: THE ACTUAL DAY OF SWITCHOVER:

Special full system backup just after midnight to TSM and Disk-to-disk (Operations)

Just after midnight there is a full system backup to TSM. Then IBM attaches the osiris SSA Disk to seter (From now on, no more access to osiris)

7:30 a.m.? Later? Earlier? Clean off seter (Joseph/Kin) Empty /sirsi, /load1, /load2, etc., i.e. empty the SCSI disk

Dan Clark (4469) or Tu switches DNS alias so that clavis and ra and osiris and seter all point to seter (Users will be using the system for public access from this second on!)

The Seter machine is given the IP address of Osiris. (Dan/Tu?)

change alias for libauth.ucalgary.ca (Roger?)

Make sure brssearch and brsvalid are in /usr/lib on seter (LP)

Reauthorize BRS databases? See scribbler. (LP)

Initialize everything on seter, do some searches (LP)

Bring up report server (it will run only the minimal list) (LP)

change .rhosts on horus to accept files from seter (for 15-minute backup of transactions)

and also allow_hosts -- see documentation page on my website for authentication.

Run daily transaction backup (needed for cash report) and cash report on seter (Nella)

Barrie restarts FRS interface, changing the domain name

Barrie restarts Clavis pin-check program

Reenable user services on seter (LP)

Halt-and-run (LP)

Watch monitor and uptime on seter (LP)

Diskless workstations????? These are OK as long as the new machine has the old domain name as one of its aliases. They point to ra.lib.ucalgary.ca

Call Roger Johnson to move the libauth pointer back to the production machine.

Call Lucille Wallace re keying manual transactions.

Email allir-m to begin using the new system for everything. (LP)

IF ANY FILE PERMISSION OR OWNERSHIP PROBLEMS: as the sirsi user, from /s/sirsi, nohup setprots >&setprots.out&

DO THE FOLLOWING ONLY IF ANY MANUAL CIRC TRANSACTIONS

Roger Johnson loads any added users

Circulation staff key manual transactions if any - depends what time the machine comes into service

Load card cancellations (LP or Roger)

in /Bincustom, run mail_catchup 02 fix_can (DISK ON SETER)
in /s/sirsi, mv forward_old .forward* (DISK ON SETER)
Last time, the weekly data from the Saturday ID update transferred over Monday morning from old server to new server. The mailfile being filled up was moved from oldserver to newserver just before Roger made the switch to start the interface coming in to newserver.
We didn't have to stop anything for the FRS interface because no acq or accounts work was done on oldserver as of Monday morning. XI10 ran on Monday night. CHECK timing with Barrie.
mail_catchup catch 02

In 2002 I ran, as sirsi in /s/sirsi,
mail_catchup can
cp forward_old .forward
mail_catchup catch

Ask IT (Angus) to switch over regular ops to seter

Restart reports on seter by renaming schedlist (Nella -- lots of details)

-------------------------------------------------------------------------

THINGS TO WATCH FOR

Remember the active crontab comes over.

The passwd.queue file needs to be reinitialized by IT

Suspend reports that you don't want running

Watch for file permission problems.

Watch for C compiler problems.

Watch for pci problems, including the pci script for start_httpd that uses the hostname as part of the path to start the Netscape webserver

Lots of steps to make sure webcat points to the new machine:

Change webcat config in the admin module. This WILL change system.env in Webcat/Config, index.html and webcat in /usr/local/etc/httpd htdocs and cgi-bin respectively.

change homepage setting in Opac Manager under Gateways

change /Webcat/Config/system.env HTTP_ADDRESS

change files in Webcat/Pages_custom/Default that refer to the other server. e.g. 54.pg, 54.pg, top.h, bclav.html, etc.

change /usr/local/etc/httpd/htdocs/index.html

change /usr/local/etc/httpd/cgi-bin/webcat

See Nella Lall's webpage on PRE-reloading tasks etc.: here

This page maintained by Linda Pearce. Last updated June 11, 2002.