Monday, 2020-03-09

« Sunday, 2020-03-08 Index Tuesday, 2020-03-10 »
*** vishalmanchanda has joined #openstack-lbaas00:00
*** threestrands has joined #openstack-lbaas00:12
*** dayou has quit IRC00:46
*** dayou has joined #openstack-lbaas00:48
dawzonWhen adding a REST field to parameters.yaml, what should I put for min_version?01:55
*** sapd1 has joined #openstack-lbaas02:14
*** ramishra has joined #openstack-lbaas03:27
*** armax has quit IRC03:49
*** sapd1 has quit IRC05:18
*** sapd1 has joined #openstack-lbaas05:31
*** sapd1 has quit IRC05:58
*** gcheresh has joined #openstack-lbaas06:12
*** sapd1 has joined #openstack-lbaas06:16
*** threestrands has quit IRC06:42
*** ccamposr has joined #openstack-lbaas06:53
*** rcernin has quit IRC07:06
*** gthiemonge has joined #openstack-lbaas07:14
*** rpittau|afk is now known as rpttau07:42
*** rpttau is now known as rpittau07:42
*** tkajinam has quit IRC08:07
*** tesseract has joined #openstack-lbaas08:12
*** rpittau is now known as rpittau|bbl08:56
*** gthiemonge has quit IRC09:16
*** gthiemonge has joined #openstack-lbaas09:30
*** gthiemonge has quit IRC09:47
*** gthiemonge has joined #openstack-lbaas09:53
*** happyhemant has joined #openstack-lbaas09:53
*** vishalmanchanda has quit IRC10:09
*** laerlingSAP is now known as laerling210:22
*** yboaron has joined #openstack-lbaas10:51
*** yamamoto has quit IRC11:29
*** nicolasbock has joined #openstack-lbaas11:37
*** yamamoto has joined #openstack-lbaas12:01
*** yamamoto has quit IRC12:06
*** yamamoto has joined #openstack-lbaas12:07
*** yamamoto has quit IRC12:29
*** takamatsu has quit IRC12:37
*** rpittau|bbl is now known as rpittau12:44
*** servagem has joined #openstack-lbaas12:45
*** yamamoto has joined #openstack-lbaas12:54
johnsomdawzon When you extend the Octavia API, you will bump the API version in https://github.com/openstack/octavia/blob/master/octavia/api/root_controller.py#L94 so whatever version your API extension is a part of would be the min_version for the docs.13:13
*** TrevorV has joined #openstack-lbaas13:19
openstackgerritAdam Harwell proposed openstack/octavia master: Support HTTP and TCP checks in UDP healthmonitor  https://review.opendev.org/58918013:36
*** yamamoto has quit IRC13:37
*** yamamoto has joined #openstack-lbaas13:44
*** sapd1 has quit IRC14:03
openstackgerritNoah Mickus proposed openstack/octavia master: WIP: Adding The abllity to set a cipher blacklist  https://review.opendev.org/71186314:16
*** yamamoto has quit IRC14:16
*** sapd1 has joined #openstack-lbaas14:17
*** armax has joined #openstack-lbaas14:20
rm_workjohnsom: err is there any way to see if the keepalived healthchecks are actually going?14:31
johnsomrm_work "going"?14:33
johnsomlike tcpdump?14:33
rm_workHmm yes I guess that's one way14:42
rm_workI was hoping for something like the haproxy HM log lines14:43
rm_workI want to see they are clearly happening and succeeding or failing14:43
rm_workAnd that if I cause the healthcheck to fail, it cycles out the member14:43
rm_workBasically, where are the keepalived logs, if any exist <_<14:44
rm_workI'm testing the patch that lets you add HTTP HMs to UDP listeners14:44
rm_workConfig LOOKS right... And the member is still up...14:45
rm_work(though the agent doesn't send anything for them in the health message, so the members still show as NO_MONITOR, will need to figure that part out)14:45
johnsomkeepalived is autonomous to the control plane.14:45
rm_workerr14:46
rm_workKeepalived that's serving the UDP members14:46
johnsomWe only capture in formation about health checks (I.e. user/tenant health) not the underlying autonomous bits14:46
rm_workNot the vrrp keepalived14:46
rm_workI'm not talking about the vrrp health14:47
johnsomAh, UDP, so that is LVS, not really keepalived. Keepalived is just a config tool for that.14:47
rm_workWe also use keepalived for UDP listeners, right?14:47
rm_workAh, ok so what I really am looking for is LVS logs14:48
johnsomLVS does not have logs really. You can get "status" information from the kernel in /proc14:48
rm_workIf such a thing exists14:48
rm_workSince it's kernel, yeah14:48
rm_workOk so I can look in /proc somewhere to see if the members are up?14:48
johnsomYes14:48
johnsomor use lvsadm which gives a pretty view14:48
rm_workAh k14:48
johnsomipvsadm14:48
johnsomsorry, always get that wrong14:49
rm_workK14:49
johnsomYeah, I thought you were talking about the actual keepalived health check scripts, etc.14:50
johnsomDifferent animal all together14:50
rm_workyeah, well the healthcheck config is in the keepalived.conf14:51
rm_workso the wording is difficult14:51
rm_workbecause it basically sounds the same i guess14:51
rm_workah, found one bug14:59
rm_workok cool, almost good15:02
rm_workjust need to fix the monitor15:02
rm_workerr, the agent reporting the monitor health15:03
rm_worknot sure honestly why it isn't working15:03
rm_work`_check_udp_listener_status` seems to look at /proc to get status for UDP members15:04
rm_workso it SHOULD be working -- this may be a larger bug?15:04
*** gthiemonge has quit IRC15:07
rm_workjohnsom: wtf, this code makes no sense to me -- I kinda get the feeling our UDP member health code is bogus? or else I'm still not understanding what's going on15:08
*** gthiemonge has joined #openstack-lbaas15:08
rm_workhttps://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/agent/api_server/keepalivedlvs.py#L239-L24415:08
johnsomrm_work I didn't write it....lol I know there are a lot of conditionals in there....15:08
rm_workbasically the only way to go ACTIVE is ...15:08
rm_workby looking in the config file and seeing if the member is there? O_o15:09
rm_workah sorry i see, that's the listener status, not the member status15:10
rm_workAHAHA ok i see where the issue is15:13
johnsomYeah, I know when I reviewed some of that, it queries some areas under /proc and parses out the operating status info15:13
rm_workyeah anyway the issue was simpl15:18
rm_workCHECKER_REGEX = re.compile(r"MISC_CHECK")15:18
rm_work->15:18
rm_workCHECKER_REGEX = re.compile(r"(MISC_CHECK|HTTP_GET|TCP_CHECK)")15:18
* rm_work blows his party kazoo15:18
johnsomlol15:19
rm_worki hate this vip-source-ip thing so much15:19
rm_workbecause it breaks healtchecks15:20
rm_workand also doesn't work for UDP lol15:20
*** sapd1 has quit IRC15:20
rm_worki always get suspicious when I make a change that fixes a bug, and run tests, and they still pass...15:27
johnsomI guess you still have coverage to add.... lol15:28
*** sapd1 has joined #openstack-lbaas15:37
rm_workmost assuredly15:49
rm_workalso found we never validated UDP HMs for single-create T_T15:50
johnsomThere is a good chance that failover will push us up to 93%15:50
rm_workso fixing that too15:50
*** gcheresh has quit IRC15:53
openstackgerritCarlos Goncalves proposed openstack/octavia-tempest-plugin master: Fix amphora list-show to run in a busy environment  https://review.opendev.org/71131616:00
*** gcheresh has joined #openstack-lbaas16:26
openstackgerritAdam Harwell proposed openstack/octavia master: Support HTTP and TCP checks in UDP healthmonitor  https://review.opendev.org/58918016:28
*** yboaron has quit IRC16:30
*** gcheresh has quit IRC16:34
rm_workok, that is looking good16:38
rm_worktesting seems to be promising in my QE cloud16:39
*** gcheresh has joined #openstack-lbaas16:43
*** nicolasbock has quit IRC16:53
*** nicolasbock has joined #openstack-lbaas16:55
*** ccamposr has quit IRC17:02
*** gcheresh has quit IRC17:13
*** rpittau is now known as rpittau|afk17:27
*** gcheresh has joined #openstack-lbaas17:28
*** sapd1 has quit IRC17:52
*** gcheresh has quit IRC18:01
*** gcheresh has joined #openstack-lbaas18:40
*** gcheresh has quit IRC18:55
*** gcheresh has joined #openstack-lbaas19:10
stevenglasfordmy slides took about 5 minutes to do, at a very reasonable speed19:14
stevenglasfordwrong chat lol19:14
*** tesseract has quit IRC19:33
*** servagem has quit IRC19:41
*** gcheresh has quit IRC19:45
*** trident has quit IRC20:57
*** trident has joined #openstack-lbaas20:58
*** trident has quit IRC21:04
*** trident has joined #openstack-lbaas21:05
*** TrevorV has quit IRC21:35
*** rcernin has joined #openstack-lbaas21:36
*** nicolasbock has quit IRC21:55
*** yboaron has joined #openstack-lbaas22:01
*** zigo has quit IRC22:13
*** zigo has joined #openstack-lbaas22:19
zigo Hi. One of my load balancer has both of its amphora in error, and it stays in PENDING_UPDATE.22:28
zigoWhat can I do to repair ?22:28
zigoWell, in fact, now it's in provisioning_status: ERROR ...22:31
*** yboaron has quit IRC22:46
*** tkajinam has joined #openstack-lbaas22:55
*** bcafarel has quit IRC23:06
johnsomRight, it will move out of PENDING when it gives up trying whatever is broken in the cloud. Options are either load balancer failover or delete and recreate23:08
*** bcafarel has joined #openstack-lbaas23:32
*** bcafarel has quit IRC23:38
« Sunday, 2020-03-08 Index Tuesday, 2020-03-10 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!