| gmann | sean-k-mooney: no, neutron is not forcing scope. it is only warning in logs when enforce_scope is disabled. if enable the it fail with 403 not warning. that is from oslo side. | 00:52 |
|---|---|---|
| opendevreview | Ghanshyam proposed openstack/nova master: Test PROJECT_ADMIN APIs with no legacy rule case https://review.opendev.org/c/openstack/nova/+/824845 | 01:33 |
| opendevreview | Ghanshyam proposed openstack/nova master: Move rule_if_system() method to base test class https://review.opendev.org/c/openstack/nova/+/824475 | 02:06 |
| opendevreview | Ghanshyam proposed openstack/nova master: Convert SYSTEM_ADMIN|READER to Admin and system scope https://review.opendev.org/c/openstack/nova/+/819390 | 02:06 |
| opendevreview | Ghanshyam proposed openstack/nova master: Server actions APIs scoped to project scope https://review.opendev.org/c/openstack/nova/+/824358 | 02:06 |
| opendevreview | Ghanshyam proposed openstack/nova master: Server actions APIs scoped to project scope https://review.opendev.org/c/openstack/nova/+/824358 | 02:09 |
| *** EugenMayer1 is now known as EugenMayer | 03:02 | |
| *** tbachman_ is now known as tbachman | 04:18 | |
| opendevreview | Merged openstack/nova master: Fill the AcceleratorRequestBindingFailed exception msg info https://review.opendev.org/c/openstack/nova/+/817326 | 04:25 |
| tkajinam | hello. I'd appreciate your attention to https://review.opendev.org/c/openstack/nova/+/824830 . we need this to fix package dependency in rdo, which is currently blocking CI jobs in some puppet module repos | 07:49 |
| bauzas | tkajinam: ack, will look | 08:03 |
| tkajinam | bauzas, thank you ! | 08:06 |
| *** akekane_ is now known as abhishekk | 08:36 | |
| bauzas | gibi: other cores, 'd appreciate a quick +W on https://review.opendev.org/c/openstack/nova/+/824830 which fixes a critical bug | 08:51 |
| bauzas | stephenfin: also, if you have time | 08:52 |
| bauzas | melwitt: stephenfin: looks like we had a regression with https://review.opendev.org/c/openstack/nova/+/824280 | 08:52 |
| gibi | bauzas: I've approved | 09:26 |
| bauzas | ta | 09:27 |
| bauzas | tkajinam: ^ | 09:27 |
| gibi | bauzas: but I would like to do a follow up to clean things up | 09:27 |
| gibi | as the dependency should be really just in the test | 09:27 |
| bauzas | gibi: I have zero context about the original patch | 09:27 |
| bauzas | but I saw we add fixtures for monkeypatching eventlet, right? | 09:27 |
| gibi | right | 09:27 |
| bauzas | as we pull nova.utils now, we also pull fixtures | 09:28 |
| gibi | it is used for the fasteners rw lock | 09:28 |
| bauzas | that's what I saw | 09:28 |
| gibi | but that is only used in test | 09:28 |
| gibi | so the rw lock wrapper can be moved back to the test code tree | 09:29 |
| gibi | or at least that is my working assumption now | 09:29 |
| tkajinam | bauzas, gibi hmmm... sorry do you mind blocking https://review.opendev.org/c/openstack/nova/+/824830 by -2 or -Workflow ? | 09:32 |
| tkajinam | I'll look into that cleaner approach. we need to update requirements in rdo but it doesn't make much sense if we revert it very soon | 09:32 |
| bauzas | tkajinam: no I guess we can move on | 09:32 |
| bauzas | or revert if needed | 09:33 |
| tkajinam | ok | 09:33 |
| bauzas | gibi: do you think you could do something before we merge https://review.opendev.org/c/openstack/nova/+/824830 ? | 09:33 |
| bauzas | I'd personnally let the fix be merged | 09:34 |
| bauzas | I've tagged the bug as critical as it's holding other projects | 09:34 |
| bauzas | and once the gate is back happy, we can sort the things up and revert what we need | 09:34 |
| bauzas | but without any rush | 09:34 |
| bauzas | agreed ? | 09:35 |
| gibi | let land the fix | 09:35 |
| tkajinam | ack. thank you for your time and thoughts, bauzas and gibi ! | 09:36 |
| gibi | tkajinam: thank you for proposing the fix | 09:40 |
| * gibi is disctracted again with meetings :/ | 09:45 | |
| opendevreview | Merged openstack/nova master: Close Glance image if downloading failed. https://review.opendev.org/c/openstack/nova/+/815347 | 10:31 |
| opendevreview | Merged openstack/nova master: Add fixtures to requirements https://review.opendev.org/c/openstack/nova/+/824830 | 11:14 |
| sean-k-mooney | fixtures should be in test-requirements not requirements but otherwise adding it makes sense | 11:35 |
| sean-k-mooney | oh it was moved that is not correct | 11:35 |
| sean-k-mooney | is see how its used https://review.opendev.org/c/openstack/nova/+/824280/1/nova/utils.py and where the dep comes form | 11:36 |
| sean-k-mooney | btu we should not use fixtures for that | 11:37 |
| sean-k-mooney | is there a patch up to remove fixutres form nova.utils. you can do exactly the same with unittest.mock | 11:40 |
| gibi | sean-k-mooney: the RW lock is only used in our test code so we move the RW wrapper in our test path and move back the fixtures deps to test-reqs | 11:40 |
| gibi | I'm about to push a patch for it | 11:40 |
| sean-k-mooney | ok | 11:41 |
| sean-k-mooney | is this a differnt RWlock then the tempest one | 11:41 |
| sean-k-mooney | im not sure why we woudl need it in nova code | 11:41 |
| sean-k-mooney | as in i dont think we need the interprocess one | 11:42 |
| gibi | sean-k-mooney: hehe, it is the same RW lock | 11:42 |
| gibi | hm, no it is not inter process | 11:42 |
| sean-k-mooney | ok just a normal rwlock | 11:43 |
| gibi | yepp | 11:43 |
| sean-k-mooney | ok that makes more sense to me i was like why have we suddenly got multiple processes in our func/unit tests :) | 11:44 |
| gibi | :) | 11:44 |
| opendevreview | Balazs Gibizer proposed openstack/nova master: Move ReaderWriterLock to the test tree https://review.opendev.org/c/openstack/nova/+/824931 | 12:35 |
| gibi | sean-k-mooney, bauzas, melwitt: ^^ | 12:39 |
| bauzas | gibi: +2s | 12:51 |
| sean-k-mooney | +1 from me also this is a better way to fix the orginal bug | 12:53 |
| *** dasm|off is now known as dasm | 13:13 | |
| sean-k-mooney | nova/healthcheck/manager.py:48:62: N310: timeutils.utcnow() must be used instead of datetime.now() | 13:50 |
| sean-k-mooney | that hacking check shoudl proably be fixed to point out it means timeutils form oslo_utils | 13:50 |
| sean-k-mooney | not the timeutils package | 13:50 |
| gibi | sean-k-mooney: good point, it is probably an easy fix | 13:57 |
| sean-k-mooney | i assume that is commign from hacking based on the N prefix | 13:58 |
| sean-k-mooney | but i have nto looked is that a nova hackign check or a global one | 13:59 |
| sean-k-mooney | nova | 13:59 |
| sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/hacking/checks.py#L209-L221 | 13:59 |
| sean-k-mooney | so ya trivial fix | 14:00 |
| gibi | N means Nova :) | 14:00 |
| gibi | afaik | 14:00 |
| sean-k-mooney | H is for genereal hacking | 14:00 |
| gibi | wondering what letter heat and horizon uses :) | 14:00 |
| sean-k-mooney | well assuming they use hacking | 14:01 |
| sean-k-mooney | mybe l for local | 14:01 |
| gibi | heat uses Heat304 for example, clever, it does not need to be a single letter | 14:04 |
| gibi | horizon uses M | 14:04 |
| gibi | sean-k-mooney, bauzas: btw, can I get eyes on this bugfix https://review.opendev.org/q/topic:bug/1952941 I'm still on the hook to backport this to victoria due to downstream upgrade failures. | 14:07 |
| sean-k-mooney | ah that one sure | 14:09 |
| gibi | thanks | 14:15 |
| sean-k-mooney | im +1 on that and the repoducer below it, thanks for following that pattern it makes it supper clear that the fix is correct | 14:16 |
| sean-k-mooney | you have a seperate change too to stop persiting the numa toloplogy in the request spec correct? | 14:17 |
| sean-k-mooney | https://review.opendev.org/c/openstack/nova/+/820215 | 14:17 |
| sean-k-mooney | after the current backportable change is done i woudl still be happy to proceed with that too | 14:18 |
| gibi | sean-k-mooney: yeah, I'm happy to move forward with https://review.opendev.org/c/openstack/nova/+/820215 if there is a consensus | 14:20 |
| gibi | the current state of that patch already show that the idea is feasible | 14:20 |
| sean-k-mooney | ya its not required but since it will prevent this type of error in the future its proably worth it in the long run | 14:21 |
| bauzas | gibi: ack, sorry was on meeting | 15:01 |
| bauzas | gibi: will look at it later today | 15:01 |
| gibi | bauzas: thanks, no worries | 15:01 |
| * bauzas was explaining the ugly sphagetti code from unshelve to Uggla :p | 15:01 | |
| bauzas | gibi: gmann: btw. if we need to update a microversion for unshelve (adding a new host), are we OK if we could remove https://github.com/openstack/nova/blob/1ddb8f83adef964a8ca050994a43adc6175994f1/nova/api/openstack/compute/schemas/shelve.py#L31 ? | 15:02 |
| bauzas | availability_zone is Optional in the doc :) | 15:03 |
| artom | Yeah, it's a weird one | 15:18 |
| gibi | bauzas: looking | 15:19 |
| artom | AZ isn't actually optional, the entire request body is | 15:19 |
| sean-k-mooney | well the az is optional | 15:19 |
| artom | If you have a body, AZ needs to be *something*, could be None | 15:19 |
| bauzas | you need to either provide no dict or if you provide the dict, you need to add the AZ key | 15:19 |
| artom | ... which could be None :P | 15:19 |
| * artom thinks we should use Uggla's unshelve to specific host RFE microversion to fix that | 15:19 | |
| bauzas | so, when adding a new argument like "host", we would also need to add AZ | 15:20 |
| sean-k-mooney | yep but there is no other property you can currently pass in the dict | 15:20 |
| bauzas | yeah, that's why I'm asking to remove this for the new API microversion also adding the new argument | 15:20 |
| bauzas | sean-k-mooney: correct, for the moment | 15:20 |
| sean-k-mooney | artom it would have to be the litral value null potentailly but this is not the only place where its slightly odd | 15:20 |
| gibi | I think both unshelve to an AZ or unshelve to a specific host make sense | 15:20 |
| bauzas | sean-k-mooney: but once Uggla will add a new key like "host", that would mean that the AZ key should be required | 15:21 |
| sean-k-mooney | yep likely you dont need to sepcify both but if you did we shoudl vlaidated the host to az relationship | 15:21 |
| artom | No...? Having to specify both host and AZ seems weird | 15:21 |
| artom | I'd say one or the other, but not both? | 15:21 |
| bauzas | that's why I'm saying we should remove the requiring when we modify this API | 15:21 |
| bauzas | in a new microversion of course | 15:22 |
| sean-k-mooney | bauzas: well as currently specified but in generally unshele:{"host":"my-host"} | 15:22 |
| sean-k-mooney | i think shoudl be valid | 15:22 |
| artom | To my mind what would be clearest is: either empty body, or AZ with a non-null value, or host with a non-null value | 15:22 |
| bauzas | anyway, let's wait Uggla to provide his spec for Z :) | 15:22 |
| sean-k-mooney | i dont think we shoudl required Az if the host is passed | 15:22 |
| gibi | artom: ++ | 15:22 |
| bauzas | sean-k-mooney: agreed, that's my concern | 15:22 |
| bauzas | sean-k-mooney: https://github.com/openstack/nova/blob/1ddb8f83adef964a8ca050994a43adc6175994f1/nova/api/openstack/compute/schemas/shelve.py#L31 doesn't accept it for the moment | 15:22 |
| bauzas | so we need to remove this line | 15:23 |
| bauzas | with a new microversion and when adding a new argument | 15:23 |
| sean-k-mooney | yes in the new microverion | 15:23 |
| sean-k-mooney | in the current one it shoudl remain | 15:23 |
| bauzas | ok, anyway, I think we have a consensus | 15:23 |
| bauzas | sean-k-mooney: of course, I know it | 15:23 |
| sean-k-mooney | yep just make sure Uggla captures it in the spec for next cycle | 15:24 |
| bauzas | I was asking whether it was ok to remove the line at the same time we add a new argument | 15:24 |
| sean-k-mooney | i assume they will start working on the implemantion in parallel | 15:24 |
| bauzas | sean-k-mooney: correct, I discussed that with him :) | 15:24 |
| bauzas | anyway, consensus, that's it | 15:24 |
| bauzas | it was just a question | 15:24 |
| bauzas | a simple one | 15:24 |
| * bauzas goes off for getting his daughter from the school | 15:25 | |
| sean-k-mooney | yep i broght that up as we have not created the new spec dir yet but Uggla can totally do that and we can review the spec | 15:25 |
| Uggla | cool, I'll work on this way. | 15:26 |
| sean-k-mooney | but ya my vote is for make it not required and supprot null as well | 15:26 |
| Uggla | passing both option, do you think that passing host should "discard" az | 15:30 |
| Uggla | passing both options, do you think that passing host should "discard" az | 15:30 |
| Uggla | I mean whatever we will have in az will be ignored if host is passed. | 15:31 |
| Uggla | I mean whatever we will have in az will be ignored if host is passed ? | 15:31 |
| gibi | Uggla: no, I think we should not ignore input. we either forbid both to be passed at once, or we should validate that the host is in the az | 15:32 |
| gibi | I more with artom, to forbid to pass both in the same request | 15:32 |
| sean-k-mooney | well either we validate that the host is in the az or it should be an error to pass both in my view | 15:33 |
| gibi | sean-k-mooney: I agree | 15:33 |
| Uggla | ok probably simpler to make it an error. | 15:34 |
| sean-k-mooney | yep simple 400 | 15:34 |
| sean-k-mooney | no existing client will pass host | 15:34 |
| sean-k-mooney | and when optinign into the new microversion they can ensure that only one is generated | 15:34 |
| Uggla | Is it possible to do this kind of exclusive or parameters in the schema validation ? | 15:35 |
| sean-k-mooney | yes | 15:35 |
| artom | Right, but why do more work when less work is enough? :) | 15:36 |
| sean-k-mooney | im pretty sure there is a way to defeint them as a mutualy exclucive group | 15:37 |
| sean-k-mooney | i just cant think of a place off the top of my head whwere we actully do that | 15:37 |
| sean-k-mooney | i suspect this si normally check after the scheme validation | 15:37 |
| Uggla | Is making both options ok in schema validation and then doing the exclusive part in the code, sounds ok ? | 15:39 |
| sean-k-mooney | https://github.com/openstack/nova/blob/1ddb8f83adef964a8ca050994a43adc6175994f1/nova/api/openstack/compute/schemas/servers.py#L172-L173 | 15:39 |
| sean-k-mooney | oneOf | 15:39 |
| Uggla | sean-k-mooney, nice thank you ! | 15:40 |
| sean-k-mooney | as i said we rearly enforce that in teh scheme but no one will object if you do it either | 15:40 |
| sean-k-mooney | there are some other exampels in the samee file | 15:41 |
| sean-k-mooney | https://github.com/openstack/nova/blob/1ddb8f83adef964a8ca050994a43adc6175994f1/nova/api/openstack/compute/schemas/servers.py#L281-L298 | 15:41 |
| bauzas | sean-k-mooney: I'll create the specs directory once we know the Z name :) | 15:42 |
| bauzas | unless I name it 'zombie' directly :p | 15:43 |
| sean-k-mooney | oh ya good point | 15:45 |
| sean-k-mooney | i still quite like "zenith" cause its all down hill from there | 15:46 |
| gmann | bauzas: gibi sean-k-mooney Uggla artom yeah, we can remove the AZ from being 'required' from schema now as it will ne extended to accepting 'host' too. AZ was added as 'required' because this API only accept None or AZ and no other field but that is handled by additionalProperties=False so 'required' is not needed as such. | 15:53 |
| gmann | bauzas: I think you will be right about 'zombie' :) everyone liking it :) | 15:54 |
| sean-k-mooney | if its is zombie we shoudl make it a base for a downstream realse | 15:56 |
| sean-k-mooney | because a decade form now we will still have once custoemr that refuses to move off it and it will never die | 15:57 |
| sean-k-mooney | like any good zombie :) | 15:57 |
| gmann | :) | 15:58 |
| sean-k-mooney | osp 18 actuly should either be the z release of a release depending on timeing so maybe that will happen anyway | 15:59 |
| sean-k-mooney | *Z release or A release | 16:00 |
| bauzas | A = Awesome release I guess | 16:00 |
| bauzas | easy peasy | 16:01 |
| sean-k-mooney | B = Buffy to help with the zombie problem ? | 16:01 |
| bauzas | C = Chilly | 16:02 |
| bauzas | D = Draughful | 16:02 |
| Uggla | I like B especially if Z is zombie ! | 16:03 |
| Uggla | Sarah Michelle Gellar was so cute. | 16:14 |
| opendevreview | Merged openstack/nova master: ensure samples folder exists for microversion https://review.opendev.org/c/openstack/nova/+/813672 | 16:15 |
| opendevreview | Merged openstack/nova master: libvirt: Ensure all volume drivers log the instance whenever possible https://review.opendev.org/c/openstack/nova/+/780260 | 17:39 |
| opendevreview | Merged openstack/nova master: functional: Add reproducer for #1907775 https://review.opendev.org/c/openstack/nova/+/766771 | 18:17 |
| ade_lee | sean-k-mooney, lyarwood hey guys, I updated the fips job to include the full set of tempest tests and there were quite a few more failures -- can you guys take a look? | 19:44 |
| ade_lee | https://review.opendev.org/c/openstack/nova/+/790519 | 19:44 |
| ade_lee | cinder folks are looking into other potential problems with volumes / cryptsetup here -- https://review.opendev.org/c/openstack/cinder/+/790535 so some of the failures may be related to that. | 19:45 |
| sean-k-mooney | sure | 19:46 |
| sean-k-mooney | have the general centos8 gate issue been fixed | 19:47 |
| sean-k-mooney | i.e. the fact they broke ping | 19:47 |
| sean-k-mooney | making it require root to run | 19:47 |
| ade_lee | sean-k-mooney, not sure .. that woud certainly muddy things | 19:47 |
| ade_lee | fungi, clarkb ^^ any idea? | 19:48 |
| fungi | sean-k-mooney: ade_lee: i don't know if red hat has fixed centos stream 8's packages yet | 19:49 |
| sean-k-mooney | the fips failrues might be unrealtaed but that is breaking all centos8 stream jobs currently right | 19:49 |
| fungi | technically, centos stream 8 was (maybe still is) broken/regressed. i recommend not relying on it since it doesn't seem like they actually test it like they did the old centos | 19:49 |
| sean-k-mooney | fungi: for what its worth i agree with not working aroudn this in our jobs | 19:50 |
| fungi | i think there's work underway to get rocky linux added, and we have openeuler though it sounds like it may have a slightly too old kernel for the fips testing | 19:50 |
| *** artom__ is now known as artom | 19:50 | |
| sean-k-mooney | fungi: really centos 8 stream should be fixed and maintianed as if it was rhel | 19:50 |
| sean-k-mooney | ade_lee: qemu-kvm: -accel tcg: mprotect of jit buffer: Cannot allocate memory | 19:51 |
| fungi | rocky or alma seems like it might be a better option if centos stream is going to be the unstable hopper where free users get to find the bugs and tell red hat what to fix before they tag a new release of their commercial product | 19:51 |
| sean-k-mooney | fungi: as someone who has to maintain said product if they dont keep centos stream stable i will continue to do all my dev on ubuntu | 19:52 |
| sean-k-mooney | i really dont like how closely tied our openstack product is to rhel or the rhel lifecycle today. if it is to contibute more value then the bruden it brings they need to keep centos 8 stream stable or make licenses avaiabel for opensrouce to sue for ci on rhel itself | 19:54 |
| fungi | but yeah, maybe this is a one-off, or maybe it'll be the kick needed to get more thorough testing in place for centos stream package updates | 19:54 |
| sean-k-mooney | if they dont using rocky or alma and not support centos makes complete sense | 19:54 |
| sean-k-mooney | fungi: we have a log runnign downstream backlog itme to get a centos stream devstack jobs running in our donwstream ci at somepoint but making it work with our patched repos is non tivial | 19:55 |
| ade_lee | sean-k-mooney, so -- the mprotect thing -- what does that mean? | 19:56 |
| sean-k-mooney | ade_lee: it looks like the host vm ran out of memory | 19:56 |
| ade_lee | sean-k-mooney, so we should be testing using larger vms? I think I recall folks having to do that somewhere -- checking .. | 19:57 |
| sean-k-mooney | ade_lee: no you need to reduce the concurancy | 19:58 |
| sean-k-mooney | ade_lee: how have you defiend the regex | 19:58 |
| fungi | unless you can get by with smaller flavors for the cirros guests or something | 19:58 |
| sean-k-mooney | the senario test need to be run serially after the other test | 19:58 |
| sean-k-mooney | fungi: the vms are using 128mb flavors | 19:59 |
| sean-k-mooney | that should be ok | 19:59 |
| sean-k-mooney | i wonder if the host vm is missing the extended swap | 19:59 |
| sean-k-mooney | that we are using to workaround the 1G tcg buffer | 20:00 |
| ade_lee | sean-k-mooney, the cinder tests are running with swap .. https://review.opendev.org/c/openstack/cinder/+/790535/24/.zuul.yaml | 20:00 |
| ade_lee | configure_swap_size: 4096 | 20:00 |
| sean-k-mooney | that proably needs to be increased | 20:00 |
| sean-k-mooney | fungi: do you know what we set it to for the debian jobs | 20:01 |
| ade_lee | sean-k-mooney, ack - well , its not set at all in the nova tests we just ran -- so we need at least that .. | 20:01 |
| sean-k-mooney | devstack allocates swap by default i belive | 20:01 |
| sean-k-mooney | 4096 for centos https://github.com/openstack/devstack/blob/3c98c21fec60da8d2d39df2e1d9b845a51817a0e/.zuul.yaml#L644 | 20:02 |
| sean-k-mooney | which is the same as debain https://github.com/openstack/devstack/commit/f8e00b86aee9a8f9646bf5aed2c618843307b963 | 20:03 |
| sean-k-mooney | im not sure if the vars section are merged | 20:04 |
| sean-k-mooney | https://review.opendev.org/c/openstack/nova/+/790519/18/.zuul.yaml | 20:04 |
| sean-k-mooney | so you might be overwriting it | 20:04 |
| ade_lee | sean-k-mooney, we made the same change in swift tests - and got better results .. | 20:05 |
| sean-k-mooney | ya you are https://0d2f8596f517931fcac8-e8ce9722d0830f321adb7d7d98f7ea38.ssl.cf1.rackcdn.com/790519/18/check/nova-centos8-stream-fips/372ef4e/zuul-info/inventory.yaml | 20:05 |
| sean-k-mooney | ade_lee: swift does not spwan vms | 20:05 |
| ade_lee | sean-k-mooney, ack | 20:06 |
| sean-k-mooney | the swap is needed because qemu started allocating a 1GB cache for tcg byty code per qemu process | 20:06 |
| ade_lee | sean-k-mooney, so try 8192? | 20:06 |
| sean-k-mooney | meanign your 128mb vms now takes 1156mb | 20:06 |
| sean-k-mooney | well right now you have 0 | 20:06 |
| sean-k-mooney | so add configure_swap_size: 4096 | 20:07 |
| ade_lee | ok so lets try 4096 .. | 20:07 |
| sean-k-mooney | you can bump it to configure_swap_size: 8192 but we also only have 80G of disk space in the vms so dont set it too large | 20:07 |
| sean-k-mooney | fungi: have we considerd using zram by the way | 20:07 |
| sean-k-mooney | in the long run that might help in general | 20:07 |
| opendevreview | Ade Lee proposed openstack/nova master: Add check job for FIPS https://review.opendev.org/c/openstack/nova/+/790519 | 20:08 |
| sean-k-mooney | lets see if that helps and we can take another look tomorow | 20:09 |
| ade_lee | sean-k-mooney, ack | 20:15 |
| fungi | sean-k-mooney: i'm not familiar with zram, is that on-the-fly memory compression? | 20:27 |
| sean-k-mooney | fungi: yes basically it creats a compressed ramdisk that you then confirure for swap with a higher priority then normal swap on disk | 20:53 |
| sean-k-mooney | with zstd compression you can get up to a 4x ratio | 20:53 |
| sean-k-mooney | fungi: fedora started using it by default for laptops and low memeory isntalls | 20:53 |
| fungi | mmm... then why not just compress the ram itself? i guess it's a simple composition to do the same, the extra layers just give me a moment's pause | 20:54 |
| sean-k-mooney | fungi: it is simpler to integrate in the kernel this way | 20:55 |
| sean-k-mooney | you can use zswap which will do it transparently but zram+swap has is more compatiable | 20:55 |
| sean-k-mooney | windows and osx just compress ram transparently | 20:55 |
| sean-k-mooney | but on linux the least intrusive way is to just add the zram device as a higher priority swap device | 20:56 |
| sean-k-mooney | https://fedoraproject.org/wiki/Changes/Scale_ZRAM_to_full_memory_size | 20:56 |
| sean-k-mooney | https://fedoraproject.org/wiki/Changes/SwapOnZRAM | 20:57 |
| sean-k-mooney | thos are the two relevent fedora propsoals | 20:58 |
| sean-k-mooney | oh ubuntu 22.04 might have it on by defaul https://www.cnx-software.com/2022/01/13/ubuntu-22-04-zswap-raspberry-pi-4-2gb-ram/ | 21:01 |
| sean-k-mooney | looks like they are going with zswap directly which also works | 21:03 |
| fungi | neat, well i expect we'll have 22.04 lts beta images available within a couple months | 21:19 |
| sean-k-mooney | the zram feature has been avaiable in the herel since 2013 so technially we coudl enable it for our existing image by just adding zswap.enabled=1 to the kernel boot line in dib | 21:20 |
| sean-k-mooney | it might be fun to play with that and see if it helps in general | 21:21 |
| sean-k-mooney | https://ubuntu.com//blog/how-low-can-you-go-running-ubuntu-desktop-on-a-2gb-raspberry-pi-4 is a nice write up on how they used it to make the 2GB rpi work | 21:21 |
| sean-k-mooney | im pretty sure the same parmaters would work well for us in our ci images | 21:21 |
| fungi | if you want to test with that, the fips mode role in zuul-jobs is an example of rebooting with custom kernel options at the start of a job | 21:31 |
| fungi | (bringing this discussion full-circle) | 21:32 |
| sean-k-mooney | oh good point. ya i might put a DNM patch on top to try enabling it and see if it helps with the job. | 21:33 |
| sean-k-mooney | if it does i could take a look at creating a dib element to configure this i guess | 21:33 |
| sean-k-mooney | the default should speed things up a little but you can tune it a little more by cahnging the algoritim and compresor values | 21:34 |
| sean-k-mooney | as always its a trade off between speed and compression ratio but lz4 and zstd are both good choices | 21:34 |
| sean-k-mooney | i think it default to lzo which is not bad but the alternitives are bettter | 21:35 |
| opendevreview | sean mooney proposed openstack/nova master: [WIP] add initial healthcheck support https://review.opendev.org/c/openstack/nova/+/825015 | 21:38 |
| sean-k-mooney | gibi: bauzas i have a lot more work todo but that is the basic infra in place more or less. | 21:39 |
| sean-k-mooney | ill start wiring it into the nova context object and such in follow up patches as well as start working testing in parallel. | 21:40 |
| ade_lee | sean-k-mooney, results are in - that worked a lot better .. | 22:03 |
| ade_lee | sean-k-mooney, https://review.opendev.org/c/openstack/nova/+/790519 | 22:03 |
| clarkb | re working around the centos 8 issue in jobs I think that might be ok as long as you understand other people may not be able to reproduce. From the CI system perspective we won't (and shouldn't) work around it for you as exposing these problems is exactly why we test | 23:00 |
| *** tbachman_ is now known as tbachman | 23:01 | |
| *** dasm is now known as dasm|off | 23:11 | |
| *** tbachman_ is now known as tbachman | 23:11 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!