- Meeting of 2nd July is cancelled and replaced by the pre-GDB on the 8th
- Presentation of Ilija of Xcache status, deployement in ATLAS and issues
- Checksum issue presented in the talk and enforce by Paul Millar : Xcache provides the opportunity to read only relevant part of files (should be the default in the future). But this prevents to check the file sanity by computing its checksum. The different source of corruption and their impact should be reviewed. This increases the criticality of the file access reliability and the report on corrupted files through degraded hardware (network connection,...).
- Operational cost to operate caching system (Paul Millar) : A priori, the failure of the disk server within the cache will only temporary affect the operation. But, if the disk server has to be reset, it will not do its role of cache and will induce an increase of network activity. The operational burden and criticality of a xcache server should be quantified.
- Ilija is following these points which will be reviewed in the pr-GDB meeting on 8th July.
There are minutes attached to this event.