opendevreview | Ke Niu proposed openstack/tooz master: remove unicode prefix from code https://review.opendev.org/c/openstack/tooz/+/854585 | 00:39 |
---|---|---|
opendevreview | Hervé Beraud proposed openstack/oslo.messaging stable/yoga: Change default value of "heartbeat_in_pthread" to False https://review.opendev.org/c/openstack/oslo.messaging/+/854904 | 08:50 |
*** brinzhang0 is now known as brinzhang | 09:33 | |
opendevreview | Daniel Bengtsson proposed openstack/pbr master: fix: check for the installed version of importlib_metadata. use it only if pbr supports it. Closes-Bug: #1972975 Change-Id: If67caac8d7ee7d5f22d0c6d262582b432d3370d1 https://review.opendev.org/c/openstack/pbr/+/841222 | 13:11 |
abishop | hberaud, damani: hi, are you the folks to talk to about getting a new etcd3gw release? | 14:05 |
damani | abishop, hi | 14:10 |
damani | abishop, yes | 14:10 |
damani | i can do it tomorrow if you can wait a bit ? | 14:11 |
abishop | damani: tomorrow would be great, thx! | 14:11 |
abishop | I think I previously mentioned https://review.opendev.org/c/openstack/etcd3gw/+/827521 is needed for things to work with recent versions of etcd | 14:12 |
sean-k-mooney | tobias-urdin: looking at https://storage.bhs.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_c6c/848338/17/check/oslo.messaging-src-dsvm-full-nats/c6c21c0/testr_results.html you have got a lot working with nats already | 14:22 |
sean-k-mooney | tobias-urdin: enough that i would be surprissed if other issue come up that would prevent you deploying this | 14:23 |
sean-k-mooney | tobias-urdin: in terms of requing and persitentce that is a grey area as we do nto depend on durabl queues today | 14:23 |
sean-k-mooney | you can optionally turn them on but its not a reuiqred feature for any deployment | 14:23 |
sean-k-mooney | on the other hand the highlevel services are built with the assumtion of a reliable transprot | 14:25 |
sean-k-mooney | so requeuing could be nice but not cirtical for the driver to be useful | 14:26 |
opendevreview | Brent Eagles proposed openstack/tooz master: Collapse relevant options in the redis driver https://review.opendev.org/c/openstack/tooz/+/855039 | 14:31 |
tobias-urdin | sean-k-mooney: yeah, i would actually argue that high availability trumps the need for durable queues if done properly | 14:39 |
tobias-urdin | but i would like to introduce retrying on the client layer, so that a message is not just sent out in the blue, implement the acknowledge part in a reply/ack queue/topic | 14:39 |
tobias-urdin | i have a little feeling that it might impact performance a bit | 14:40 |
tobias-urdin | as you can see by that test job, it took some minute(s) extra for nats | 14:40 |
sean-k-mooney | tobias-urdin: our job to job test execution time is not that stable in the gate | 16:18 |
sean-k-mooney | tobias-urdin: so unless that partern continues for many jobs i would not worry yet | 16:19 |
sean-k-mooney | tobias-urdin: i would expect nats to have simialr or bertter performace vs rabbit at scale but perhaps not | 16:20 |
sean-k-mooney | the marketing hype would make me hope so but i have never actully deployed it | 16:20 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!