Tuesday, February 16, 2010

Upcoming Gratia Maintenance

Over the next two weeks, the FNAL-provided OSG Gratia services will be migrated to new hardware. The following will occur:

1) On Thursday, 2/18 data collection and DB updates will be suspended at 09:00 CST (15:00 UTC) to allow a copy of the live DB. Reports will still be available at this time, but obviously will not include data not received. This procedure will take 4-6 hours, during which time data will be backlogged by probes and sent when service resumes. Interested system administrators may note a build-up of files under $VDT_LOCATION/gratia/var: this will decrease once more when contact is re-established with the collector.

2) On Thursday, 2/25 at 09:00 CST (1500 UTC) the migration will occur. Since this will involve a migration of service IPs to the already set-up and operating new systems, the outage for each service is expected to be under 5 minutes barring unforeseen circumstances. Any probes attempting to contact during this time will backlog the data and re-send as before upon resumption of service.

The services affected by this change are:
- GRATIA-OSG-PROD (gratia-osg-prod.opensciencegrid.org:80/443)
- GRATIA-OSG-ITB (gratia-osg-itb.opensciencegrid.org:80/443)
- GRATIA-OSG-DAILY (gratia-osg-daily.opensciencegrid.org:80/443)
- GRATIA-OSG-TRANSFER (gratia-osg-transfer.opensciencegrid.org:80/443)

Each service mentioned above will gain its own new reporting URL by appending -reports to the address, viz:
- gratia-osg-prod-reports.opensciencegrid.org/
- gratia-osg-itb-reports.opensciencegrid.org/
- gratia-osg-daily-reports.opensciencegrid.org/
- gratia-osg-transfer-reports.opensciencegrid.org/
The standard URLs will continue to work due to service redirection but for efficiency reasons, please update your browser bookmarks to use the new URLS for the reporting GUI.

In addition, we propose to retire the unused "GRATIA-OSG-ITB-PERM" ("permanent ITB") reporting-only service at this time (2010/02/25 09:00 CST). There are no known users of this service. This was originally instituted to take replication of a subset ("permanent ITB") of the data received by the GRATIA-OSG-ITB service but to the best of our knowledge has never been used for same, and has anyway been dormant (available but not receiving data) for several months now without complaint. Since the service is unused, we expect the impact to be zero. People who wish the service to be resumed should notify us via GOC prior to 2010/02/25.

We do not expect any impact whatsoever at the, "Production" level: reporting to WLCG will take place as normal and we expect the out-of-date reporting on 2/18 and the momentary outages on 2/25 to be no more than a minor inconvenience to a very few people.

Please note especially: RSV services will *NOT* be affected by this migration.