Changes between Initial Version and Version 1 of developers-meeting-20140910


Ignore:
Timestamp:
09/10/14 16:46:37 (5 years ago)
Author:
karsten
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • developers-meeting-20140910

    v1 v1  
     1[[TOC(depth=0)]] 
     2 
     3[part of a [wiki:developers-meetings series of meetings]] 
     4 
     5= Participants = 
     6Gerd, Al, Paul, Tigran, Karsten 
     7 
     8= Agenda = 
     9 
     10[see box on the right-hand side] 
     11 
     12= Postcards = 
     13 
     14Up to two minutes (uninterrupted) per person where they can answer two questions: 
     15 *  What I did last week (since the last meeting), 
     16 *  What I plan to do in the next week. 
     17 
     18No questions until we get through everyone :) 
     19 
     20Karsten: Parallel reading of files from dCache 
     21 
     22Al: Posted some patched, only few missing to commit the whole thing 
     23    Stress testing the alarms, looks satisfactory, but they are only  
     24    meant for 2.11 
     25 
     26Dmitry: Mostly opertional stuff, stuck transfers, globus online 
     27 
     28Gerd: Executive meeting, writing board report, ... 
     29      Worked on migration module 
     30      operational issues, debugging handshake timeouts 
     31      debugging SRM semi-deadlock 
     32 
     33Paul: GridKA School, dCache workshop + talk about federated identity 
     34      talked about future directions, preparing 
     35      fixing releases 
     36 
     37Tigran: JGlobus 2.1 + dCache, more complicated than expected 
     38        Problem with Voms-API, send fix to JGlobus, but no response, yet. 
     39        Undid changes in dCache, adapted to new BouncyCastle 
     40        Had problems on dcache-cloud, ncount got out of sync,... investigation ongoing 
     41        Got some patches from external contributors 
     42 
     43= Plans for patch-releases = 
     44 
     45Should we make a new patch release? 
     46 
     47We want to get releases out, but currently some unit tests are failing 
     48Jenkins slaves died, etc. 
     49 
     50Investigating. 
     51 
     52Could this be because of increased concurrency and/or slowed down machines 
     53because of changed Virtual Machines state. 
     54 
     55There is a patch to remove race-condition from tests. 
     56 
     57But there are more. 
     58 
     591. AtomicCounter -> Gerd 
     602. BasicTest -> Tigran 
     613. following Tests will be distributed via eMail 
     62 
     63= Trunk activity = 
     64 
     65Progress with new features... 
     66 
     67== questions to tigran == 
     68 
     69Bulk deletetion through SRM: Atlas showed some slides with 30Hz deletion rate 
     70from multiple clients. Gerd managed to archive single client performance of 188Hz 
     71 
     72Atlas doing small file stages: Small Files scripts might help, but they are not  
     73ready to be distributed. 
     74 
     75== Globus Online == 
     76What's the status? 
     77 
     78Will be able to run tests on a production system next week. 
     79Currently there is no culprit that causes the failures. 
     80 
     81 
     82== jGlobus == 
     83 
     84We need a new jGlobus release, but unit tests are failing. 
     85 
     86Should we move away from jGlobus? We have some dependencies (principals), 
     87but other libraries provide similar functionality. Also srmClient depends. 
     88 
     89Gerd: We could have different versions for what is awailable in the dists. 
     90 
     91As soon as VOMS people upgrade, we get the new BouncyCastle and also Canl. 
     92 
     93 
     94= Issues from [FIXME: Add link to yesterday's Tier-1 meeting] = 
     95 
     96Xavier, Marc, Gerd:  
     97Xavier discovered overloading pools. Advised to upgrade pool manager.  
     98  - Billing file logs: Pool manager sending information to billing 
     99 
     100Gerd: using 2.10.3 except for some problems 
     101 
     102= Outstanding RT Tickets = 
     103 
     104[This is an [wiki:TicketActions auto-generated] item.  Don't add items here directly] 
     105 
     1068442: This should rather be logged at INFO or DEBUG and change the output default pattern 
     107      The "suspended" case, there may be an error code missing 
     108 
     1098427: Could this be caused by different root paths? -> get more config 
     110      => We should improve the error message as well. 
     111      => avoid logging it twice. 
     112 
     113= Review of RB requests = 
     114 
     1157183 -> Paul to have a look 
     1167289 -> Karsten to have a look 
     117 
     118Also please have a look at Al's Alarm patches 
     119 
     120= DTNM = 
     121 
     122Proposed: same time, next week.