Following ongoing consultation with many involved parties, an approximate-but-moderately-firm migration plan is now in place for removing anchor from service, and replacing it with newer/faster equipment.
The plan is outlined below.
In general,
- Deleting data is not in any way part of the plan.
- All files presently on "anchor" will remain intact, and will be accessible "transparently" to end users on the "new deployed replacement anchor"
- If you have problems with code running on the new anchor, the "old anchor" will remain accessible for at least one month, to help debug and migate any problem code.
- note that access to "old-anchor" will be only granted on an as-required basis; by default all accounts will be locked on old-anchor to encourage people to use the "new anchor"
- User accounts which have been dormant on anchor for an extended period will be LOCKED on the "new anchor" as a preventative security measure. If you believe your account has been inappropriately locked, and you require access - please contact Tim Chipman to resolve the problem promptly.
- In general, active user accounts are available to "present and/or active" members of oceanography department, as per the POLAN general guidelines / terms of service. (Please see http://seahorse.phys.ocean.dal.ca/wiki/pmwiki.php/Polan/Policy for details)
ANCHOR MAIL USERS
(This is of interest only to those folks who use "Pine" on anchor to read / send emails)
- During the migration day (Monday Nov.5th) you will not have access to anchor, nor email that is stored locally on anchor / accessed via PINE.
- You will, as always, be able to use the webmail interface running on the mail server, to access new/inbox email and any other mail stored on the mail server itself.
DATES AND TIMELINE:
- Tuesday, Oct.30th: Cargo is retired from service, 3511 disk array with 'acoustic' data is migrated to 'newanchor'
- Tuesday, Oct.30th: Initial User Data Synchronization from Anchor to NewAnchor is started (should be complete in <72hours)
- Wednesday, Oct.31st: Cargo is fresh-installed with CentOS 5.0/64-bit and SunRay (thin-client-workstation) server software
- Monday, Nov.5th: Inbound realtime data for LB project stops arriving on Anchor
- Monday, Nov.5th: Anchor goes offline; all data becomes inaccessible temporarily. Final synchronization of data from Anchor->Newanchor begins (takes <6 hours (*))
- Monday, Nov.5th: Deploy SunRay thin client workstation server
- Monday, Nov.5th: Deploy "Anchor" replacement server; final validation; then open for service
- Tuesday Nov.5th-> Thursday Nov.8th: Assorted debugging and tweaks on an "as-required" basis for the deployment of the new anchor.
(*) CLARIFICATION ON TIMELINE OF ROLLOVER DAY - MONDAY NOV.5th
- Once anchor goes offline, and final data synch begins to "newanchor" - other steps must wait before proceeding
- the data synch will most likely be quite a lot faster than 6 hours. It will likely be 1hr or less, but I can't be certain
- once data sync is complete, other steps will follow (turning on sunray server, bringing up the new "anchor" - hopefully complete by lunchtime, but no certainty here.
- note that other servers / workstations which NFS mount home directories (such as JADE) **WILL** experience service / problems while homedirs are offline. This is unavoidable with this type of work.