1. Reading part of the tape challenge finished last week. Results look promising -- expected throughput was 1.93GB/s, we achieved ~ 1GB/s more than this.
2. There was a major LHCb Dirac update on Monday, which introduced some issues. Recovered within several hours. There were lot of failed jobs due to this.
3. Low number of running LHCb jobs due to insufficient number of production requests.
4. Consistency check identified some dark and lost data. Dark data was removed, lost files were re-replicated by (all data operations were done by the LHCb Computing team).

Tickets:

1. Slow checksums (stats):
   • Still waiting
2. Deletion problems
   • Solved
3. Problems with simultaneous access to the same file on ECHO
   • On hold, tests are ongoing at Glasgow
4. Vector read.
   • One more test: what happens with the LHCb applications is vector read requests returns "wrong" (i.e. not the one that was requested) data. This was tested (the same patch, but once in a 1000 vector reads it shifts one of the requested chunks by 1 byte), and it seems like the application crashes.
   • Dedicated patched WN for production LHCb jobs is being prepared.