Attending: Sam, Jens (chair+mins), John B, Marcus, Steve, RobC, Winnie, David, Dan, Raul, Govind, Matt Apols: Simon (in ATLAS meeting), Brian (on hols) 0. Operational blog post. Yes, we have blog posts! Thank you, Marcus! ... any other things from last week? Accounting revisited - dCache, DPM, and CASTOR "wrong" but in different ways; and there is no general means of publishing accounting data from StoRM even if the data is generated via df or du -s or something equivalent. Related to accounting, we have poked glue-wg for updates to the GLUE-for-ATLAS questions (from November - http://storage.esc.rl.ac.uk/weekly/20161116-minutes.txt) Marcus' problem was an iptables problem. Marcus also asked about the MySQL settings, suggesting that splitting tables into different files could increase performance, or at least reduce the volume. Previously he'd had a 20GB file, and with tables split they reduced to 7-8GB in total [although that could have been the reimporting compacting the tables?] One could perhaps see if it's a good idea by comparing loads on the system. There is a need for cleanups, not just for requests but also with ATLAS, RUCIO famously creates lots of empty directories. Another note on the xrootd configuration; the settings required to support VOMS properly - another case for documenting in the wiki... Govind notes about his test (prior to putting it into production) system, he seems to be the first person in the UK to try DPM on CentOS7. Experiences suggest tweaking is necessary, and BDII is not working. However, Andrea is quite responsive, so it may be worth us having a chat with him (Andrea) and decide whether to go forward with CentOS7 or to downgrade. Sam notes a boundary case with DPM deletions through the DAV interface, where errors are reported due to a check which checks the file after it has been deleted - so a message is created but the deletion has done the right thing. 1. ... no one seems to have taken the hint to update the storage KeyDocs (not even myself ...) - still needed, though. Also a reminder to Everyone, that while each KeyDoc has an owner, Everyone should look at the storage documents and update them if needed. 2. Hardware testing discussion - Simon's thing and what we suggest he do with it. And how we can collaborate on the testing. Documenting tests on web page? It's "only" access to the system, not a loan as such - so we can only access it and run tests with it; it may be more difficult to change the RAID setup or install DPM on it, or any such useful or interesting thing. Also interesting would be the performance of a degraded system, including the time to rebuild/recover. But there are still interesting things to test, cf Marcus' script, including different file sizes. Presumably XMA would also want something from it, by giving us access to this. DDN got a whitepaper. It would be worth clarifying - what sort of outcome XMA want - what's the scope of the test - whether it is just raw performance or we can add stuff - how we do it, like if anyone else wants to work with Simon on it. 3. AOB Marcus Ebert dpm.nohv1 Marcus Ebert dpm.listvoms raul We should be able to sample the disk servers raul The use ionode or fio to build a test case raul But, of course, SImon must already be on it John Bland what was the link for that again? Samuel Cadellin Skipsey https://wiki.egi.eu/wiki/APEL/Storage John Bland thanks Daniel Traynor du -sh ./*/ Daniel Traynor it might take a while to run