Monday, 2014-08-25

*** ptoohill has joined #openstack-lbaas00:29
*** Zebra has joined #openstack-lbaas02:19
*** woodster_ has joined #openstack-lbaas02:34
*** amotoki has joined #openstack-lbaas04:29
*** sbfox has quit IRC04:47
*** sbfox has joined #openstack-lbaas04:54
*** orion_ has joined #openstack-lbaas05:59
*** orion_ has quit IRC06:04
*** jschwarz has joined #openstack-lbaas07:37
*** sbfox has quit IRC07:40
*** enikanorov has quit IRC08:26
*** enikanorov has joined #openstack-lbaas08:26
*** Zebra has quit IRC10:34
*** mestery has quit IRC13:10
*** mestery has joined #openstack-lbaas13:17
*** mestery_ has joined #openstack-lbaas13:21
*** mestery has quit IRC13:22
*** orion__ has joined #openstack-lbaas13:23
*** orion__ has quit IRC13:24
*** orion__ has joined #openstack-lbaas13:24
*** sballe has joined #openstack-lbaas13:43
sballeMorning. Back from vacation :-)13:59
sballeblogan do you know if any of the LBaaS were accepted? Mine didn't get in14:00
jschwarzhey sballe14:02
sballejschwarz, hi14:02
*** amotoki has quit IRC14:14
*** enikanorov__ has joined #openstack-lbaas14:27
*** pckizer_ has joined #openstack-lbaas14:28
*** electrichead has joined #openstack-lbaas14:31
*** pckizer has quit IRC14:32
*** clev has quit IRC14:32
*** enikanorov_ has quit IRC14:32
*** redrobot has quit IRC14:32
*** clev has joined #openstack-lbaas14:33
*** dlundquist has joined #openstack-lbaas14:41
*** orion__ has quit IRC14:43
*** ptoohill has quit IRC14:43
*** orion__ has joined #openstack-lbaas14:44
*** orion__ has quit IRC14:47
*** clev has quit IRC14:50
*** clev has joined #openstack-lbaas14:50
*** dlundquist has quit IRC14:52
*** dlundquist has joined #openstack-lbaas14:55
a2hillsballe, "Load balancing as a service v2.0 - Juno and beyond" seems to have made it in15:00
a2hillnot sure of any others though15:00
*** sbfox has joined #openstack-lbaas15:00
sballea2hill, that is great news!15:04
*** electrichead is now known as redrobot15:04
*** mestery_ is now known as mestery15:06
a2hillIndeed it is, still waiting on final words as if we will actually be able to make it or not15:07
*** mikedillion has joined #openstack-lbaas15:09
*** markmcclain has joined #openstack-lbaas15:10
*** dlundquist has quit IRC15:16
*** jschwarz has quit IRC15:16
*** xgerman has joined #openstack-lbaas15:16
*** sbfox has quit IRC15:18
*** markmcclain has quit IRC15:19
*** dlundquist has joined #openstack-lbaas15:21
*** markmcclain has joined #openstack-lbaas15:21
*** sbfox has joined #openstack-lbaas15:22
*** samuelbercovici has joined #openstack-lbaas15:35
*** sballe_ has joined #openstack-lbaas15:35
*** sballe has quit IRC15:37
*** markmcclain has quit IRC15:46
*** sballe_ has quit IRC16:01
*** samuelbercovici has quit IRC16:01
*** sballe_ has joined #openstack-lbaas16:01
*** markmcclain has joined #openstack-lbaas16:15
*** dlundquist has quit IRC16:15
*** markmcclain has quit IRC16:20
*** mlavalle has joined #openstack-lbaas16:25
*** sbalukoff has joined #openstack-lbaas16:28
sbalukoffMorning folks!16:34
*** markmcclain has joined #openstack-lbaas16:37
*** dlundquist has joined #openstack-lbaas16:45
*** openstackgerrit has quit IRC17:00
*** openstackgerrit has joined #openstack-lbaas17:13
*** mikedill_ has joined #openstack-lbaas17:15
*** mikedillion has quit IRC17:19
*** crc32 has joined #openstack-lbaas17:21
*** markmcclain has quit IRC17:25
*** crc32 has quit IRC17:25
*** blogan has quit IRC17:31
*** TrevorV has quit IRC17:32
*** a2hill has quit IRC17:32
xgermandid I count right: - jenkins is the tie breaker?17:33
*** crc32 has joined #openstack-lbaas17:35
sbalukoffSince it's a tie, my thought is to try doing the meetings via IRC for a couple weeks and then re-vote in IRC. This gives us a good comparison of how the meetings go (and how much we cover) in IRC versus via voice...  and after a couple weeks, if some of those who voted aren't showing up to the IRC meetings they voted for... well... you've got to be here for me to care about your vote. :)17:49
rm_workI agree with sbalukoff because it gets me what I want :P17:49
rm_work(at least temporarily)17:49
sbalukoffWho knows. Maybe some people who voted for video meetings may change their vote, too.17:50
xgermanlike jenkins!17:52
*** sbfox has quit IRC17:56
rm_workhmm, flying to Seattle Wednesday night17:57
*** sbfox has joined #openstack-lbaas17:59
dougwigwhat's the count among people that actually attend?18:01
crc32whats the link. I think the voting is intentionally being left open just long enough for the IRCers to take the lead.18:01
*** dlundquist has quit IRC18:02
crc32sbalukoff: We all know what the IRC experience is already.18:02
crc32speaking of. My team is having issues connecting to IRC now actually. lol18:02
dougwigcrc32: dude, no one is intentionally trying to force IRC via voting tactics.  uncool.18:03
crc32we already agreed on webex meetings. Then a vote goes into a blue print. no need to tell me whats uncool.18:04
rm_workalso, if there was a clear indication that "leaving the voting open longer" leads to "more people voting for IRC" then … wouldn't that just be an indication that more people want IRC? lol18:04
crc32When are you comming into work adam?18:05
crc32we have stuff to talk about.18:05
rm_worki said in internal-irc that i was prolly going to just WFH since nothing I'm working on relies on being in the office or even talking to anyone on the team right now :P18:05
rm_workwhat do we have to talk about?18:05
rm_workI wasn't aware of anything18:06
crc32Are you in tomarrow?18:06
dougwigsbalukoff: last time i added it up, among actual attendees, it was pretty strongly in favor of webex.  i'm not seeing numbers that would say we should actually switch to IRC yet.18:06
crc32ok. Well talk then.18:06
dougwigcrc32: if you ignore the objections of all those that couldn't voice their opinion in the very medium that got voted for, sure.18:07
crc32dougwig. Let it go. I'm not arguing for the meetings to be neither here nor there anymore.18:08
dougwigstop insinuating nefarious motives and i'll let it go.18:08
xgermanlet's meet by chain letter18:08
crc32dougwig: It was a joke but what ever. This is why I don't like IRC. You don't get to see me smiling. And I'm assuming your smiling too. But well never know.,18:09
dougwighah, fuck, got me. i'm sorry.18:10
xgermanyou know about emoticons18:10
rm_workcrc32: lol, half the time i can't tell you're joking IRL either :P18:10
crc32rm_work: If my voice is high pitched is pretty obvious. And if you see me using straw man arguments its supposed to be an attempt at humour.18:10
dougwigmy humor detector is greatly dimished when i have to wake up super early and get on airplanes.18:11
dougwigdminished, even.18:11
rm_workdougwig: are you still plane-ing basically every day?18:11
crc32Its a good thing no one jokes on the red line to mosow. (The red phone was really just a serial link with text)18:12
dougwigno, just every few weeks.  out to california today18:12
crc32adam is your bouncer working. The team is having IRC issues.18:13
crc32Don't know how I got in.18:14
rm_worknot sure18:15
rm_workI am not on VPN18:15
rm_workcrc32: looks like it is offline18:15
crc32ok that explains it.18:15
rm_workcrc32: let me go fix it18:15
crc32I've been having to join via I've noticed going via DNS resolution is a fail for me18:16
rm_workthat server is on iNova18:18
rm_workmy bots are down too18:18
rm_workI wonder if the iNova cluster went down18:18
rm_workor if they shut it off18:18
rm_workthat'd be super annoying18:19
rm_workI had a ton of stuff configured on there that took me a while to set up <_<18:19
*** sbfox1 has joined #openstack-lbaas18:20
*** sbfox has quit IRC18:20
*** sbfox1 has quit IRC18:20
*** sbfox has joined #openstack-lbaas18:20
rm_worknot sure if i missed anything, or if my text went through, but18:22
rm_workcrc32: looks like my iNova server with the bots and the ZNC is down18:22
rm_workcrc32: not sure I have the credentials to that iNova account anymore -- I just borrowed them briefly from jorgem18:22
crc32ok back18:29
crc32had to deal with an auto claim.18:29
rm_workhmm, wtf inova18:29
rm_worktrying to find an email with those creds but i think it was never in an email anyway18:30
rm_workand if it were, it would have expired long ago from my inbox18:30
crc32theres a yellow light on the stick. Let me see what went down.18:30
crc32edir borkage.18:33
crc32oh well18:33
crc32plus firmware to switches was pushed.18:33
rm_workcrc32: do you have access to that inova account?18:33
rm_workto list the servers on it / see status?18:33
crc32Looks like theirs a fire alarm in the ORD datacenter. It looks like they evacuated.18:34
crc32just kidding18:34
crc32well if eDirs is having issues I'm guessing iNova would be the last to recover.18:35
crc32I don't do I nova. I use CloudServers directly.18:35
crc32CloudServers is less stingy.18:35
rm_workyeah but they aren't inside VPN :/18:36
rm_workwe can move this discussion to internal IRC I think :P18:36
crc32Thats the great part about it. Your not restricted by the VPN anymore.18:36
dougwigi was enjoying this window into rax dirty laundry.18:38
crc32its really dirty.18:47
crc32I'm telling them they should use nylon.18:47
crc32its too flamable18:47
crc32err shoulden't use nylon18:47
crc32hey sbalukoff did you get a letter saying if your talk was accepted or not?18:52
*** blogan has joined #openstack-lbaas19:15
crc32hey brandon19:15
crc32Did you come in on the bouncer or on your own?19:15
bloganno bouncer19:15
bloganbouncer got bounced!19:16
rm_worki'm still tracking this thing down19:20
*** mestery has quit IRC19:20
rm_workdunno wtf happened but i am betting it's iNova's fault19:20
*** mestery has joined #openstack-lbaas19:20
blogani blame the series of tubes19:25
crc32I blame your mother.19:25
blogani read that in sean connery's voice19:25
rm_workthey're bouncing the instance now, hopefully it'll come back up fine19:25
bloganso i can see issues arising on this webex/irc vote19:29
crc32yea its called "Who are these people?"19:30
bloganwho's that crc32 guy19:30
crc32thats Carlos Diablo Garza. Senior software developer for Lbaas.19:30
bloganDiablo? really19:31
crc32He also sits right behind you.19:31
crc32Yes I had it legally changed like Dragon.19:31
crc32I was going for another word but was convinced to just go with Diablo.19:31
bloganoh yeah me too, Brandon En Feugo Logan Junior, Esquire, III19:32
*** pckizer_ is now known as pckizer19:37
*** mikedill_ has quit IRC19:41
sbalukoffNote to all app developers out there:  Do not write your web app such that your client aggressively attempts to re-connect without an exponential back-off if the back-end is serving requests slowly.19:47
sbalukoffThat's one idiotic customer-created crisis (temporarily) fixed.19:48
sbalukoffcrc32: It looks like none of the talks ideas I submitted or was included in got accepted.19:48
sbalukoffDid anyone else's talks here get accepted?19:48
crc32Are you still going sbalukoff. I'm doing a stock sale to finance my trip. :(19:49
blogansbalukoff: did you get a talk accepted?19:49
sbalukoffcrc32: I'm pretty sure I still am. Need to clear that with my bosses first, but the chances that we're going to be (at least informally) meeting and collaborating on Octavia stuff pretty-much guarantees it.19:50
sbalukoffblogan: No, none of the 3 talks I submitted got accepted.)19:50
xgermanyeah, Alex gave me his spot19:50
xgermanso Susanne and I will be there19:51
xgermanwhile Alex remains in beautiful Seattle...19:51
sbalukoffI will say I'm tempted to take a trip to Boise at the same time as the Paris summit though: The election is happening right in the middle of it.19:51
sbalukoffSeattle in November...  ok, yes, it can still be beautiful here, if the gray hasn't settled in yet. ;)19:52
xgermanyeah, and Paris doesn't have an ocean...19:52
*** blogan_ has joined #openstack-lbaas19:54
xgermansbalukoff you were planning to go door to door and canvass?19:54
sbalukoffOn the whole IRC vs. webex meeting thing: dougwig: Yes if you count the number of votes from regular meeting attendees the numbers definitely favor keeping the meetings in webex.  However, if we as a group are going to be (defacto) accused of not being open, or not doing things the openstack way, especially by people influential to OpenStack in general like Steffan...  I'm tempted to spend a couple of weeks in IRC, just a19:54
sbalukoffs proof that those voting weren't actually interested in contributing. If they then come back at incubation candidate time and point to this and say "See, you don't do things the OpenStack way" we can absolutely call bullshit on that.19:54
*** ptoohill has joined #openstack-lbaas19:54
xgermanstart with dougwig's home - I hear he is undecided :-)19:54
*** TrevorV has joined #openstack-lbaas19:54
sbalukoffxgerman: Heh! I might end up doing that when I visit my folks (probably in September).19:54
sbalukoffxgerman: Ha!19:55
sbalukoffxgerman: Me showing up at his door might convince him to vote for the other guy. XD19:55
xgermantrue. Not sure how they feel about out of state influencers19:55
sbalukoffxgerman: Or some guy who looks like the comic book shop owner from the Simpsons standing on their doorstep.19:56
dougwigsbalukoff: I can get behind that plan.  and +1 to visiting19:56
sbalukoffcrc32: And yes, I think we do have enough experience with the IRC stuff to know how it goes. I would be surprised if we can cover even 1/3rd of the topics we cover in the webex meetings. :P19:58
crc32for me the order goes. 1. Face to face. 2. Video conference 3. Teleconfrance. 4. IRC  5. Email 6 Telegraph 7. Snail male 8. Hand flag semaphore 9. Smoke signals20:00
crc3210. Pheramones20:00
rm_workcrc32 / blogan: everything is back up20:00
rm_workyou can log in to the bouncer now20:00
crc32I alerted them.20:00
rm_workcrc32: can't underestimate those pheramones20:00
rm_workant pathing with pheramone trails is awesome20:00
crc32you ever play with ants like that. Run your finger across their path and watch them get lost. Then you time them to see how long they re converge their network.20:01
blogansbalukoff: i was hoping that fast and the furious talk would have got accepted20:03
blogansbalukoff: i didn't think the octavia talk would ahve gotten much traction since no one knows what it is20:04
crc32selling guns and racing cars again blogan?20:04
sbalukoffblogan: I was hoping all the talks would have been accepted. I think we all had great ideas to share, eh.20:04
blogancrc32: with vin diesel20:04
crc32Or the dead guy.20:04
blogansbalukoff: well I'm glad only one of mine got accepted, I don't like public speaking20:04
blogan_oh dear god bouncer just connected20:05
sbalukoffWell, there's definitely still the possibility of having an Octavia design session. I'm actually more interested in seeing that happen than seeing an Octavia talk happen.20:05
blogan_me too20:05
sbalukoffblogan: which talk of yours got accepted?20:05
*** blogan has left #openstack-lbaas20:05
blogan_the neutron lbaas one20:06
sbalukoffCool beans.20:06
blogan_it wasn't MY talk, it was just a talk I was in20:06
blogan_i suspect it will be a talk on "this is what we have planned. Questions?" "INCUBATOR? WHEN WHY WHERE HOW?"20:07
rm_workcrc32: Paul Walker RIP T_T20:14
rm_workso, I would still go to the summit if RS paid for me to go, but no way am I going to pay my own way20:15
rm_workI'll just have to make sure blogan_ stays on IRC via mobile or something20:15
rm_workso we can have a good communication channel :P20:15
rm_worksbalukoff: or make you promise to get on IRC whenever possible20:15
*** mikedillion has joined #openstack-lbaas20:23
xgermansbalukoff Octavia design session = goodness20:26
xgermanwe should figure out how to submit talks for that20:26
rm_worksad than most of our team won't be going to Paris :/20:27
xgermanalso -- for some reason the two process setup deosn't perfrom as well20:27
blogan_rm_work: ill be on webex the entire time20:27
rm_workblogan_: lol >_<20:27
rm_workthanks blogan_… I think.20:27
blogan_it'll be nothing but my face walking around20:28
blogan_selfie videos!20:28
rm_workah man, I skipped lunch today <_< whoops20:28
*** sbfox has quit IRC20:32
openstackgerritTrevor Vardeman proposed a change to stackforge/octavia: Initial creation of db models, modules, and tests
blogan_TrevorV says: douche20:38
*** sbfox has joined #openstack-lbaas20:41
blogan_you shoudl totally leave comment questions asking why he made a decision to do a certain thing20:42
sbalukoffxgerman: Could you help me understand how you come to that conclusion in the numbers below? Also, it doesn't look like the tests were actually run evenly:  One of them used "-c 500" and the other "-c 400"21:07
sbalukoffI'm hoping that was a simple oversight--  we want this to be an apples-to-apples comparison, eh.21:07
xgermanwell, I ran with -c 500 and it caused errors21:08
xgermanso I dropped it to the value which didn't cayse errors21:08
xgermanBut I am not happy with those values... will liley play around some more21:10
sbalukoffThat's fine, I guess-- but you'll want to drop to "-c 400" for the first test too, then.21:10
xgermanthe first test can handle -c500 without errors21:10
xgermanI haven't figured out how to make ab scale up and tell me what the maximum sustainable concurrency is21:11
sbalukoffFrom my perspective, unless the sections are labeled incorrectly, it looks like the second one (two processes, one listener each) actually performs better. (Higher req/sec, faster data throughput, etc.)21:11
sbalukoffRight...  so on our load balancers there's quite a bit we do in /etc/sysctl.conf to tune the kernel for high-performance TCP.21:11
xgermanI went pretty much stock...21:12
sbalukoffIf the haproxy processes, or the servers serving the requests don't have this tuning, one can run into bottlenecks related to poor network stack tuning. ;/21:12
sbalukoffSo, haproxy should be able to serve on the order of 10,000+ requests / sec from a single process in any case.21:12
sbalukoffNot being able to do that points at a tuning problem, IMO. :)21:13
xgermanwell, that worries me, too -- but on the other hand Octavia is usppsoed to use random novavms21:13
sbalukoffI can forward to you what we put in the /etc/sysctl.con if you'd like.21:13
xgermanthat would be helpful21:13
sbalukoffThat's true-- but these nova VMs will probably have this tuning. (Might not help if the underlying kernel used by the hardware isn't tuned, but we can only do so much, eh.)21:14
rm_workyeah, I believe in our deployment at least, the plan was to be running on a cluster that was specifically tuned for our application21:15
rm_worknot just "generic public cloud VMs"21:15
xgermanwe were thinking generic cloud vms for now21:16
xgermanbut if benchmarks suggest that plan is not feasible we will rethink21:16
xgermanthat's why I am running it there :-)21:16
rm_workheh, yep21:17
sbalukoffxgerman: Ok, forwarded you our usual TCP tuning parameters.21:18
xgermanthanks --21:18
xgermanI will try those21:18
sbalukoffAlso wanted to say:  Loading any kernel connection tracking modules is not good for this kind of high performance proxy. They'll cause slow-downs and lost packets, and really aren't necessary given what the proxy is doing.21:18
sbalukoff(ie. if 'lsmod | grep conntrack' shows anything, you're going to have a bad time.)21:19
sbalukoffI realize it may not be possible not to have this on the underlying host... but we definitely don't need it on the VM.21:19
*** mikedillion has quit IRC21:21
dougwigstock sysctl is tuned for a desktop.  it sucks for a server.21:22
sbalukoffdougwig: Indeed!21:24
*** markmcclain has joined #openstack-lbaas21:30
*** enikanorov_ has joined #openstack-lbaas21:57
*** enikanorov has quit IRC21:58
*** johnsom_ has joined #openstack-lbaas22:09
*** xgerman has quit IRC22:11
*** xgerman has joined #openstack-lbaas22:12
*** HenryG_ has joined #openstack-lbaas22:12
*** johnsom has quit IRC22:13
*** HenryG has quit IRC22:15
*** sballe__ has joined #openstack-lbaas22:16
*** sballe_ has quit IRC22:19
*** markmcclain has quit IRC22:21
*** sbfox has quit IRC22:38
xgermanwell, error rare remains slightly higher with tow processes -- but otherwise there is no major difference after the runing22:40
rm_workin other words, the tuning helped?22:41
*** fnaval has joined #openstack-lbaas22:49
*** sbfox has joined #openstack-lbaas22:59
*** markmcclain1 has joined #openstack-lbaas23:04

Generated by 2.14.0 by Marius Gedminas - find it at!