I agree its important that we test and understand the behavior a typical user will experience. And fix it if it's not what we want. Typical user? ==? running on a laptop on a so so network. ;-) Sent from my iPhone
On Jan 4, 2014, at 19:43, "Brettin, Thomas S." <[email protected]> wrote:
I think having a small team, maybe led by Shane, that looks into this would be great. A team of two really would do it. One to run a ramped DoS sort of code, and another to monitor the stack (mysql , mongo, star man, nginx, memory, network gear, etc)
Having this done in parallel to Bob's changes. If Bob can roll the changes in early, say as soon as he arrives, then all deployments would have the decaying retry capability.
t
On Jan 4, 2014, at 7:38 PM, "Rick @ Argonne" <[email protected]> wrote:
Tom
As you know I'm also seeing reliably issues with some servers/services. Perhaps we need a dive on that. Bob thinks a modification to the client side code generator can support a smart retry that should improve things and be reflected in all clients. I'm also wondering if we have something configure to limit number of sever connections per client to foil denial of service attacks that might be causing some of this. ?
Sent from my iPhone
On Jan 4, 2014, at 19:15, "Brettin, Thomas S." <[email protected]> wrote:
Right. When I was writing the messae, I was realizing there was going to be a phase two, someting beyond the 15th. As I was writing I wondered if i should discuss the two phases in the same email. What I describe is what we'll do, but not everything that we will do of course.
Yes, the idea to whiteboard things is good. This will elicit the info needed to fully understand what the 'private' deployment will be. I need Dan and Shane present for that.
Can we commandeer a deployment war room with a whiteboard, or is there no such room available? Having the small conference room just for the deployment team, with whiteboards with modules, dependencies, readiness, exceptional variations, etc would be useful.
On Jan 4, 2014, at 6:56 PM, Adam Arkin <[email protected]> wrote:
Hi Tom ,
I wish we could enforce this... but in my speaking to the various teams I know there are going to be issues with meeting this deadline. The question right now is how to deal with this is an orderly way given that what you voice is the theoretically correct way...
Search won't complete because not everyone has their types or data in the system yet, Landing pages and the attendant widgets are still in development, WSS is not complete for a variety reasons, and there are a few other elements that are not all there.
I propose we need to, in the first couple of days next week, take over a big board in the conference room and try and get all the modules and their interdependencies onto the board, determine the elements ready and not for deployment, and then prioritize and assign names to the tasks for completion.
Some links for docs that might be useful: Narrative: http://kbase.us/docs/narrative/functions/index.html
Workspace object format: https://docs.google.com/a/lbl.gov/document/d/1cFaSLxtW7QfsIhlreRko_AE_nRoLJ2...
Documentation of Widget making and landing pages:
https://docs.google.com/a/lbl.gov/document/d/1Jwy9uIet1HoJcqjOgvyjYpJyFlIKVK...
This is a spreadsheet of indexed objects and whether or not they have landing pages: https://docs.google.com/a/lbl.gov/spreadsheet/ccc?key=0AvN8PTXC3N8wdEVXdlBBL...
So-- we need some way to phrase your message that says we need to get as much deployed by the 13th as possible and we will be accounting for and scheduling those elements that cannot make it because of missed dependencies.
Sigh...
A
On Sat, Jan 4, 2014 at 4:36 PM, Brettin, Thomas S. <[email protected]> wrote: Here is a draft announcement for the schedule for the deployments. Of course, the hard deadline for deployment to production is Jan 15.
Builds are scheduled to take place from Jan 7-13, 2014. Activities will cease on the end of Jan 15 and at that time we will take inventory of what made it into production. The build pattern is mostly unchanged, module owners communicate to the release team modules that are ready. The release team deploys the module to the test environment, executes the tests, and then deploys the module to production. Triage based on failures occurs along the way, and module owners have the opportunity to fix problems and resubmit the module as ready to the release team. The release team will take no new deployment requests after Jan 13, so that the Jan 15 production deployment schedule can be kept. On Jan 16, a report to Adam will be compiled by the release team and submitted.
On Monday, Jan 6, or as soon as a the release team arrives, the KBase services spread sheet will be finalized, and the build punchlist will be reviewed.
Feedback welcome. I will send tomorrow or later this evening if Adam you are ok with this schedule.
Tom
On Jan 3, 2014, at 1:19 PM, Adam Arkin <[email protected]> wrote:
That would be great. We need to get a hard end to deploy. Do we have a plan for a public vs. "private" instance?
A
On Fri, Jan 3, 2014 at 11:05 AM, Thomas Brettin <[email protected]> wrote: I can do a formal timeline for prod deploys. Would formally start on the 13th. of Jan.; after the hackathon, but we can start earlier on those that are ready.
Would work out the logistics of tracking with Miriam, Nomi, Dan, Shane next week
On Jan 3, 2014, at 12:58 PM, "Brettin, Thomas S." <[email protected]> wrote:
I can do a timeline for prod deploys. Would formally start
Sent from my Verizon Wireless 4G LTE DROID
Adam Arkin <[email protected]> wrote:
Hi All,
I know already people are pushing the deadline... I am very worried about this. If we are going to push the deadline we should do so formally and set the process for final deployment and testing.
We also were going to have a public site and a site with more advanced functionality for friends. Is this going to happen?
Should we get people on a schedule? How should we manage this?
A
-- Adam Paul Arkin Dean A. Richard Newton Memorial Chair ------------------------------------- Director, Physical Biosciences Division E.O. Lawrence Berkeley National Laboratory
Professor, Department of Bioengineering University of California Berkeley, CA, 94720
CEO/CSO, DOE Systems Biology Knowledgebase, http://kbase.us Director, Berkeley Synthetic Biology Institute, http://synbio.berkeley.edu PI and Co-Director, ENIGMA SFA, http://enigma.lbl.gov Investigator, Energy Biosciences Institute, http://energybiosciencesinstitute.org
Office: 512C Energy Biosciences Building (Berkeley Campus) Mailing address: E.O. Lawrence Berkeley National Laboratory 1 Cyclotron Road, MS 955-512L Berkeley, CA 94720
Contact: W: http://genomics.lbl.gov V: 510-495-2366 C: 510-206-1389 F: 510-486-6219
Assistant: Gwyneth A. Terry V: 510-495-2116 E: [email protected] -------------------------------------
-- Adam Paul Arkin Dean A. Richard Newton Memorial Chair ------------------------------------- Director, Physical Biosciences Division E.O. Lawrence Berkeley National Laboratory
Professor, Department of Bioengineering University of California Berkeley, CA, 94720
CEO/CSO, DOE Systems Biology Knowledgebase, http://kbase.us Director, Berkeley Synthetic Biology Institute, http://synbio.berkeley.edu PI and Co-Director, ENIGMA SFA, http://enigma.lbl.gov Investigator, Energy Biosciences Institute, http://energybiosciencesinstitute.org
Office: 512C Energy Biosciences Building (Berkeley Campus) Mailing address: E.O. Lawrence Berkeley National Laboratory 1 Cyclotron Road, MS 955-512L Berkeley, CA 94720
Contact: W: http://genomics.lbl.gov V: 510-495-2366 C: 510-206-1389 F: 510-486-6219
Assistant: Gwyneth A. Terry V: 510-495-2116 E: [email protected] -------------------------------------
-- Adam Paul Arkin Dean A. Richard Newton Memorial Chair ------------------------------------- Director, Physical Biosciences Division E.O. Lawrence Berkeley National Laboratory
Professor, Department of Bioengineering University of California Berkeley, CA, 94720
CEO/CSO, DOE Systems Biology Knowledgebase, http://kbase.us Director, Berkeley Synthetic Biology Institute, http://synbio.berkeley.edu PI and Co-Director, ENIGMA SFA, http://enigma.lbl.gov Investigator, Energy Biosciences Institute, http://energybiosciencesinstitute.org
Office: 512C Energy Biosciences Building (Berkeley Campus) Mailing address: E.O. Lawrence Berkeley National Laboratory 1 Cyclotron Road, MS 955-512L Berkeley, CA 94720
Contact: W: http://genomics.lbl.gov V: 510-495-2366 C: 510-206-1389 F: 510-486-6219
Assistant: Gwyneth A. Terry V: 510-495-2116 E: [email protected] -------------------------------------