|Main||EGI.eu operations services||Support||Documentation||Tools||Activities||Performance||Technology||Catch-all Services||Resource Allocation||Security|
WARNING: The meeting is being moved from 20th of June to 20th of June, because 13th of June is holiday in Netherlands and other countries. Sorry for the inconvenience.
Detailed agenda: Grid Operations Meeting 20 June 2011 14h00 Amsterdam time
1. Middleware releases and staged rollout
1.1 EMI-1 release status (Cristina)
1.2. EMI/UMD current status
1.3. Staged Rollout (Mario)
1.3.1 gLite 3.1 series
- LFC 1.8.0-1, patches in staged rollout state with no EA - Decision was to close those patches since there was no interest shown from production sites, and the support calendar has ended.
1.3.2 gLite 3.2 series
- CREAM 1.6.6 and glite-SGE_utils are under staged rollout - fix a bug seen in the previous staged rollout
- L&B 2.1.21: staged rollout done, report produced - possibly will be released together with cream and sge.
1.3.3 EMI1 - UMD1
- Requesting at OMB tomorrow, a delay of one week for the UMD official annoucement, from 4 July to 11 July.
- Staged rollout tests with date of 23 June have been done or near finalization, reports already received, other expected until 28 June.
- L&B and LFC_mysql verification reports are expected today and tomorrow, and will go to staged rollout next.
- MPI rejection: a new version has already been produced in EMI, and will be released soon.
- WMS: rejected due to several problems found in verification. One is fixed in the EMI update of last week, but there are still others that need to be fixed - proxyrenewal in particular. Discussing with EMI if we can have all this fixed soon so that it can still enter in the SW provisioning process and be released in UMD1.
- StoRM: will be released in an EMI update this week.
- UNICORE-Registry: rejected due to a technical mistake in the SW provisioning process (not due to a bug), will be released in the next update.
1.4 Interoperability (Michaela)
2. Operational Issues
2.1 Requirement #1388
NGI_IT requested documentation about guidelines on how to migrate a service preserving its state (e.g. an SE), on how to set up hot-failover for stateful services.
- Instruction on what is needed to be copied and preserved.
- Instruction on how to set up failover or redundant instances of the same service.
NGI_IT provided parsed EMI-1 products documentation and highlighted the components that are missing documentation on this issues.
2.1.1 Services that do not need documentation
The following services already report in their documentation the procedures to migrate the service maintaining the state, and a description of an high availability configuration.
- APEL publisher (HA not docuemented but not critical)
- VOMS-Oracle replica not documented
- Stateless service, documentation about critical log files to save/maintain
- Failover and load balancing available for both server/client side
- HA not documented, but WMS can be configured to point to multiple LB
2.1.2 Services that miss documentation
|Service||How to migrate the service||How to set up failover load balancing||Comments|
|LFC||?||?||Documentation available for both HA and migration. But the docs may be not up-to-date. The load balancing tecnique based on DNS round robin is missing.|
|BDII||V (not applicable)||X||Service stateless, no documentation available about HA and load balancing|
|DPM||V||?||The DPM architecture is modular, but deployment scenarios are not well covered in the documentation|
|CREAM||?||X||Cream is a stateless service, but for accounting and security purpose, some logs must be preserved to be compliant to EGI recommendation. It is not clear which files are important and which are not.|
2.1.3 Other middleware stacks?
NGI_IT gathered information only about gLite components, where they have more experience. All the EMI services will be deployed from scratch.
Are the same informations needed also for ARC components? The interested NGIs could parse the available documentation and report the component that miss information.