OSG Technology Area Meeting, 27 August 2018¶
Coordinates: Conference: 719-284-5267, PIN: 57363; https://www.uberconference.com/osgblin
Attending: BrianB, BrianL, Carl, Derek, Edgar, Marian, Marco Mambelli, Mat
- 2018-09-03 meeting canceled (Labor Day)
- TimT OOO until Thursday
- Suchandra OOO until Wednesday afternoon
- This week: Mat
- Next week: Suchandra
- 5 (-4) open tickets
|# of tickets||Δ||State|
|4||+3||Ready for Testing|
|0||+0||Ready for Release|
OSG Software Team¶
- AI (Suchandra, TimT): Update your Y7 effort numbers
- AI (Derek): sugwg-scitokens.phy.syr.edu is uploading pilot records is unregistered in the topology
- A few submit hosts are uploading pilot records. Why? How does the Gratia probe decide what kind of record to upload?
- AI (Carl, Edgar): Review the user support Gratia configuration
- AI (Mat): request GlideinWMS 3.4.0 promotion so that Marco can test it
- Arizona State (BrianL, Suchandra): consistently running 240 pilots through their hosted CE. Reverse DNS still an issue for their own CE.
- Cyverse (Mat) - aragon submit host topology registration
- LIGO (Edgar) - performance diffs between reading lots of small files vs big files of the same total size. They are seeing 10x the number of read failures for the small files case
- Harvard (BrianL): still having issues with multicore pilots and their CE
- UC Davis (BrianL): mixed CE and backend HTCondor configuration. Documentation updated to clarify the difference between a CE and backend HTCondor.
- UFL (BrianL, Edgar): xrootd-lcmaps still doesn't respect the –policy option
OSG Investigations Team¶
- XRootD documentation and docker overhaul is ongoing. Organized in StashCache in collaboration with Slate and NRP (PRP) collegues.
- Shepherding the xrootd plugins through the release process.
- xrootd-TPC - contrib
- xrootd-macaroons - contrib
- xrootd-scitokens - release
- x509-scitokens-issuer - contrib
- Discussions with with Internet2 to place StashCache in connection points. Waiting on GeoIP issues to be worked out.
- (Still ongoing) The repo server is going to move, once again. But there should be no noticable differences. If there is, then we failed.
- Topology-itb server certificate failed last week. Fixed by Mat, but underlying issue is that puppet is failing on topology-itb, or turned off.
- rsync service on the repo hosts have been puppetized but still need to be added to checkmk. UNL admins to manually verify rsync functionality after the repo move.
- AI (BrianL): Give BrianB a link to the XRootD use case doc