opendevreview | Takashi Kajinami proposed openstack/ironic master: Remove unnecessary egg_info options https://review.opendev.org/c/openstack/ironic/+/904053 | 02:35 |
---|---|---|
opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent master: Remove unnecessary egg_info options https://review.opendev.org/c/openstack/ironic-python-agent/+/904054 | 02:37 |
opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent master: Remove deprecated pbr options https://review.opendev.org/c/openstack/ironic-python-agent/+/904055 | 02:42 |
opendevreview | Takashi Kajinami proposed openstack/networking-baremetal master: emove deprecated pbr options https://review.opendev.org/c/openstack/networking-baremetal/+/904058 | 02:44 |
opendevreview | Takashi Kajinami proposed openstack/networking-baremetal master: Remove deprecated pbr options https://review.opendev.org/c/openstack/networking-baremetal/+/904058 | 02:44 |
opendevreview | paresh proposed openstack/ironic stable/2023.1: [Backport] Fixes Raid creation in iLO6 and other BMC with latest schema https://review.opendev.org/c/openstack/ironic/+/904069 | 06:28 |
opendevreview | paresh proposed openstack/ironic stable/2023.2: [Backport] Fixes Raid creation in iLO6 and other BMC with latest schema https://review.opendev.org/c/openstack/ironic/+/904070 | 06:29 |
opendevreview | paresh proposed openstack/ironic stable/zed: [Backport] Fixes Raid creation in iLO6 and other BMC with latest schema https://review.opendev.org/c/openstack/ironic/+/904071 | 06:44 |
opendevreview | paresh proposed openstack/ironic master: [Docs] Adds changes to 'iLO' docs regarding Gen11 support https://review.opendev.org/c/openstack/ironic/+/904075 | 07:47 |
rpittau | good morning ironic! o/ | 08:27 |
opendevreview | paresh proposed openstack/ironic master: [Docs] Adds changes to 'iLO' docs regarding Gen11 support https://review.opendev.org/c/openstack/ironic/+/904075 | 09:05 |
opendevreview | paresh proposed openstack/ironic master: [Docs] Adds changes to 'iLO' docs regarding Gen11 support https://review.opendev.org/c/openstack/ironic/+/904075 | 09:15 |
opendevreview | Riccardo Pittau proposed openstack/ironic master: Fix versions in release notes https://review.opendev.org/c/openstack/ironic/+/904103 | 09:22 |
opendevreview | Riccardo Pittau proposed openstack/ironic-python-agent master: Fix versions in release notes https://review.opendev.org/c/openstack/ironic-python-agent/+/904104 | 09:24 |
opendevreview | Riccardo Pittau proposed openstack/ironic-inspector master: Fix versions in release notes https://review.opendev.org/c/openstack/ironic-inspector/+/904105 | 09:27 |
opendevreview | Merged openstack/sushy master: Add release version to release notes https://review.opendev.org/c/openstack/sushy/+/903984 | 09:27 |
Nisha_Agarwal | dtantsur, hi | 10:15 |
Nisha_Agarwal | dtantsur, i was looking in the sushy code ...and commenting out this line sushy makes it work seamlessly for HPE Compute Scale-Up 3200 servers | 10:44 |
Nisha_Agarwal | https://opendev.org/openstack/sushy/src/branch/master/sushy/connector.py#L53 | 10:44 |
iurygregory | good morning Ironic | 11:00 |
dtantsur | Nisha_Agarwal: hi! This is really strange.. Do you have any ideas what that happens? | 11:03 |
Nisha_Agarwal | I had put pdb... | 11:03 |
Nisha_Agarwal | and i saw when sushy gets the session token when AccessError is recvd, after that sushy session reauthenticates itself | 11:04 |
dtantsur | Does it mean that the token is only valid for one connection? | 11:05 |
Nisha_Agarwal | during this authentcation itself, session token is invalid | 11:05 |
Nisha_Agarwal | i read on the net and see that "Connection: close" means that the session should close after that call | 11:06 |
Nisha_Agarwal | after that HTTP request call | 11:06 |
dtantsur | Yep, but not the Redfish session | 11:06 |
dtantsur | Hence my question if the token is only valid for one (TCP) connection | 11:06 |
Nisha_Agarwal | requests library does the HTTP connection only right? | 11:06 |
dtantsur | yep | 11:06 |
dtantsur | So, Connection:close means that a TCP connection will be established for each request | 11:07 |
Nisha_Agarwal | dtantsur, this is something we dont know...atleast in curl we dont hit this issue | 11:07 |
Nisha_Agarwal | Yes | 11:07 |
Nisha_Agarwal | or u can say it is very very shortlived | 11:07 |
dtantsur | yeah.. curl, of course, also established a new connection | 11:07 |
dtantsur | have you tried sending "Connection: close" with curl? | 11:07 |
Nisha_Agarwal | because very few times it goes and is able to get the next call data | 11:07 |
Nisha_Agarwal | dtantsur, i tried but couldnt reproduce the issue... | 11:08 |
Nisha_Agarwal | but this is for sure that just commenting this line makes it work for all Superdome series of server | 11:08 |
Nisha_Agarwal | and we dont hit the issue even once | 11:09 |
dtantsur | Nisha_Agarwal: could you talk to your hardware colleagues about it? I'm curious to hear their reasoning. | 11:09 |
Nisha_Agarwal | dtantsur, yep....Its HPE shutdown time ....so most likely we will hear from them next year | 11:11 |
dtantsur | okay | 11:11 |
Nisha_Agarwal | still checking if anyone is theer | 11:11 |
Nisha_Agarwal | there* | 11:12 |
Nisha_Agarwal | i was just thinking that if we dont give "Connection: close" but instead "keep-alive: <time>" | 11:13 |
Nisha_Agarwal | will that help in achieving what we want to achieve using "connection: close" header | 11:13 |
dtantsur | I don't believe there is such a thing on the client side? | 11:14 |
dtantsur | (there are TCP settings but that requires control over the socket) | 11:15 |
Nisha_Agarwal | Connection management | 11:15 |
Nisha_Agarwal | Connection | 11:15 |
Nisha_Agarwal | Controls whether the network connection stays open after the current transaction finishes. | 11:15 |
Nisha_Agarwal | Keep-Alive | 11:15 |
Nisha_Agarwal | Controls how long a persistent connection should stay open. | 11:15 |
Nisha_Agarwal | This i have copied from https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers | 11:16 |
dtantsur | ah, a separate header | 11:17 |
Nisha_Agarwal | Yes | 11:17 |
Nisha_Agarwal | Connection: Keep-Alive, Keep-Alive: timeout=5, max=1000 | 11:17 |
Nisha_Agarwal | I havent tried it though | 11:18 |
Nisha_Agarwal | just a thought that this might work | 11:18 |
Nisha_Agarwal | and this may acheive the purpose of having "Connection: close" in code | 11:18 |
dtantsur | I"m still trying to understand if this header can be used in a request | 11:19 |
Nisha_Agarwal | hmmm | 11:20 |
iurygregory | what I've found by searching a bit was " allows a single TCP connection to remain open for multiple HTTP requests/responses. " | 11:21 |
iurygregory | so maybe it would work | 11:21 |
iurygregory | "http request Connection: Keep-Alive instead of Connection Close in requests" | 11:21 |
dtantsur | yeah, but all examples so far point at it being returned by a server, not set by a client | 11:21 |
dtantsur | (it's not about the Connection header, it can be used for sure) | 11:22 |
iurygregory | oh gotcha | 11:22 |
dtantsur | My worry is ending up with something that we think works, but actually is a no-op | 11:25 |
iurygregory | dtantsur, if you have a chance, can you provide input in https://review.opendev.org/c/openstack/ironic-python-agent/+/902012 ? | 13:21 |
iurygregory | I will probably work on it next week so I want to have some feedback if the procedure is ok | 13:22 |
dtantsur | putting on the queue | 13:37 |
iurygregory | ty | 13:39 |
opendevreview | Verification of a change to openstack/ironic master failed: Test redfish with reduced sushy-tools feature set https://review.opendev.org/c/openstack/ironic/+/903361 | 15:14 |
iurygregory | =O | 15:23 |
opendevreview | Verification of a change to openstack/ironic master failed: Test redfish with reduced sushy-tools feature set https://review.opendev.org/c/openstack/ironic/+/903361 | 15:36 |
JayF | dtantsur: would there be a concern around using keepalives changing performance semantics on conductors? | 16:18 |
JayF | dtantsur: Generally speaking though, this all seems like hpe's bug; I don't understand how this session handling works on any other machine we've tested it on, but not here, and it's somehow our bug (?) | 16:18 |
JayF | dtantsur: essentially I agree with your insinuation that there's something yet-to-be-unearthed here | 16:18 |
dtantsur | JayF: yeah, I'm not fond of changing that line. Even more than about conductor performance, I'm worried about overloading BMCs | 16:23 |
rpittau | good night! o/ | 16:57 |
opendevreview | Merged openstack/ironic master: Test redfish with reduced sushy-tools feature set https://review.opendev.org/c/openstack/ironic/+/903361 | 18:13 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!