OWG Meeting 23-Mar-2016
Location and Time
Room: CC F326-327
You can connect using BlueJeans Video conferencing (ID: 120 390 084). (Click "Expand" to the right for details -->):
(if problems, call phone in conference room: 757-269-6460)
- To join via Polycom room system go to the IP Address: 184.108.40.206 (bjn.vc) and enter the meeting ID: 120390084.
- To join via a Web Browser, go to the page https://bluejeans.com/120390084.
- To join via phone, use one of the following numbers and the Conference ID: 120390084
- US or Canada: +1 408 740 7256 or
- US or Canada: +1 888 240 2560
- More information on connecting to bluejeans is available.
- gluon48 -> rebuilt with RHEL6
- Gluon work disk (installed and available as /gluonwork1)
- gluon01 -> RHEL6 (summer)
- gluon46 will remain at RHEL7
- Compiler upgrade (gcc 4.8 or 4.9)
- ROOT -> v6
- Event/Data rate + IB network config
- ROL status (SYNC events)
- L3/event tagging (event_size)
- New meeting time? (-> 2pm?)
- Recharge Wednesday: Hot Fudge Sundae
Attendees: David L.(chair), Sergey F., Simon T>, Vardan G., Dmitry R., Bryan M., Sean D., Dave A., Carl T., Hovanes E., Naomi J.
- We discussed the sluggishness of the gluon cluster at times. It was suggested that this may be due to the 1GB link from gluonfs1. David will ask Paul about adding gluonfs1 to ganglia so we can monitor it's network usage.
- gluon01 will be reverted to RHEL6 in the summer
- This summer the offline group will move to requiring a c++11 compliant compiler. This will likely be gcc 4.8 or 4.9
- The online uses sim-recon for monitoring plugins (used with RootSpy) and hdview2 so we will need to have such a compiler available.
- It is unknown at this point whether significant issues will develop in compiling controls or other software with a different compiler. If that turns out to be the case, we will need to adjust the plan then.
- The offline group will likely start supporting (requiring?) ROOT6 after this upgrade. Online will then need to react accordingly by installing ROOT6 and possibly making it the default.
- Dave A. summarized the situation with gluonraid ib configurations. Both computers are now using datagram method which gives faster throughput. A bad cable was replaced on gluonraid2 which also brought it up to 13.5GB/s
- Hovanes reminded us that there was an issue with some computers not mounting the RAID disks.
- Sergey noted that he'd seen where gluonraids were not cross mounting one another. David L. said this was not always the case and may be a remnant of the issue Hovanes brought up
- David L. will check what computers currently have the gluonraids mounted and follow up with Paul on what the situation is.
ROL status (SYNC events)
- Sascha has implemented SYNC events to readout TS scaler information and check for pedestal drifts
- Two different pieces of code now parse this in the offline (one will eventually be removed)
- There is currently no checking done during sync events that the modules are actually empty. Sergey and David L. will discuss this with Sascha.
- David L. showed a few plots indicating that the event size grows linearly with beam current
- Some indications that disentangling time goes linearly with event size
- Profiling has yielded contradictory and often confusing results on why disentangling/parsing is taking so long. Further study is needed
New meeting time
- Meeting time will be moved to 2pm in the future. Due to the ongoing run, we may not expect to meet again for 6 weeks
- One remote participant objected to our advertising "Hot Fudge Sundae" since they could not partake from CMU
- There was no local support for removing advertising of the recharge Wednesday