| opendevreview | Alistair Coles proposed openstack/swift feature/mpu: mpu: refactor MPUSession class https://review.opendev.org/c/openstack/swift/+/934737 | 11:12 |
|---|---|---|
| opendevreview | Alistair Coles proposed openstack/swift feature/mpu: mpu-auditor: audit sessions https://review.opendev.org/c/openstack/swift/+/931517 | 11:17 |
| opendevreview | Merged openstack/swift feature/mpu: mpu: add func tests using an UTF8 mpu name. https://review.opendev.org/c/openstack/swift/+/934076 | 12:22 |
| opendevreview | Merged openstack/swift feature/mpu: mpu: refactor MPUSession class https://review.opendev.org/c/openstack/swift/+/934737 | 13:11 |
| fulecorafa | Hey guys, I'm having problems starting swift here. I just run swift-init restart main and it does not start the servers with "Exception: worker did not start normally: b''", didn't find other logs. Any ideas? | 13:59 |
| DHE | depending on the distro, check if logs got shoved into some swift-specific files/dir? | 14:09 |
| DHE | for troubleshooting, you can run the services manually as the swift user, typically with "swift-$type-server /etc/swift/$type-server.conf" and see stderr messages. | 14:10 |
| fulecorafa | So, no idea why, but seems the container name was it? | 14:15 |
| fulecorafa | I did the same thing multiple times with a host name as "swift-lock". Changed to "swift-why" (little frustrated, sry hehe) and it worked | 14:16 |
| fulecorafa | Nah, scratch that, didn't work | 14:17 |
| fulecorafa | I'll try initializing the server isolated | 14:17 |
| DHE | no error message doesn't give much to go on... I would normally assume bad config files or missing ring files. something like that.. | 14:19 |
| fulecorafa | So yeah, no shot here. Tried running the failing servers manually, no error message apert from the one I already sent | 14:38 |
| DHE | as the swift user? | 14:39 |
| fulecorafa | Yep | 14:40 |
| DHE | umm.. that was my only real idea... short of much more invasive methods like strace... | 15:36 |
| timburke | fulecorafa, maybe try running with workers=0? that should get us out of the spawn-a-bunch-of-workers path that makes things trickier to debug | 16:28 |
| timburke | i would've hoped that logs would still be going out via syslog :-/ | 16:29 |
| fulecorafa | "getpwnam(): name not found: 'swift'" does this help? timburke | 16:41 |
| timburke | fulecorafa, ah! so that's coming from this call in drop_privileges: https://github.com/openstack/swift/blob/2.34.0/swift/common/utils/__init__.py#L971 | 16:46 |
| timburke | this was running in a container, right? you might need to copy (or mount) /etc/passwd in the container | 16:46 |
| timburke | or run with user=root | 16:47 |
| fulecorafa | I'm not using the latest swift though | 16:47 |
| fulecorafa | I'm at commit 19879831272b37b5e49f73a4d49f1a4587ccd7c7 (HEAD, tag: 2.29.2) | 16:48 |
| fulecorafa | There's a single file for utils | 16:48 |
| fulecorafa | Also, the /etc/passwd file already exists and has reasonable configurations | 16:49 |
| timburke | it's been like this a looong time https://github.com/openstack/swift/blob/1.2.0/swift/common/utils.py#L453 | 16:49 |
| timburke | does it have a 'swift' user? | 16:49 |
| fulecorafa | That's what I though | 16:51 |
| fulecorafa | The user in the VM is called ubuntu and all configuration files point the user as ubuntu | 16:51 |
| fulecorafa | ubuntu@swift-why:/etc/swift$ grep -nr ubuntu | 16:52 |
| fulecorafa | proxy-server.conf:5:user = ubuntu | 16:52 |
| fulecorafa | object-server/1.conf:5:user = ubuntu | 16:52 |
| fulecorafa | internal-client.conf:2:user = ubuntu | 16:52 |
| fulecorafa | container-server/1.conf:5:user = ubuntu | 16:52 |
| fulecorafa | container-reconciler/1.conf:2:user = ubuntu | 16:52 |
| fulecorafa | account-server/1.conf:5:user = ubuntu | 16:52 |
| fulecorafa | object-expirer.conf:2:user = ubuntu | 16:52 |
| fulecorafa | Greping swift reveals nothing | 16:52 |
| fulecorafa | But the egg:swift parts | 16:52 |
| timburke | huh. weird -- the user in drop_privileges should be coming out of the configs: https://github.com/openstack/swift/blob/2.29.2/swift/common/wsgi.py#L461 | 16:56 |
| timburke | not sure why it's looking for 'swift' if you've got 'ubuntu' configured | 16:58 |
| fulecorafa | Huh, juts went there. It's falling in the base case | 16:59 |
| fulecorafa | As in it's not finding the config user... | 17:00 |
| fulecorafa | This should be my guide, thx a lot timburke | 17:00 |
| timburke | sure -- let me know how it goes | 17:01 |
| fulecorafa | Trying to understand now why it is not capturing this information | 17:02 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Deprecate expirer options https://review.opendev.org/c/openstack/swift/+/920452 | 17:02 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Configurable expiring_objects_task_container_per_day https://review.opendev.org/c/openstack/swift/+/933373 | 17:02 |
| opendevreview | Clay Gerrard proposed openstack/swift master: refactor: remove iter_task_accounts_to_expire https://review.opendev.org/c/openstack/swift/+/919639 | 17:02 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Parallel distributed task container iteration https://review.opendev.org/c/openstack/swift/+/918366 | 17:02 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Delete s3api MPU segments when expiring the manifest https://review.opendev.org/c/openstack/swift/+/800701 | 17:02 |
| fulecorafa | There we go! I was missing the user in part of the files. They got overwritten in some commits | 17:04 |
| fulecorafa | Thanks a lot for the help guys <3 | 17:04 |
| timburke | nice -- fwiw, i think we often put user in the [DEFAULT] section | 17:09 |
| fulecorafa | Yep. It so happens that someone pushed an overwrite that we didn't expect | 17:10 |
| opendevreview | Tim Burke proposed openstack/swift master: DNM: investigating probe test timeouts https://review.opendev.org/c/openstack/swift/+/934350 | 17:39 |
| opendevreview | Tim Burke proposed openstack/swift master: probe tests: Set default timeout for subprocesses https://review.opendev.org/c/openstack/swift/+/934813 | 17:39 |
| opendevreview | Shreeya Deshpande proposed openstack/swift master: Remove PrefixLoggerAdapter and SwiftLoggerAdapter https://review.opendev.org/c/openstack/swift/+/931060 | 21:09 |
| opendevreview | Tim Burke proposed openstack/swift master: probe tests: Run relinker via subprocess https://review.opendev.org/c/openstack/swift/+/934910 | 23:39 |
| opendevreview | Clay Gerrard proposed openstack/swift master: add backoff sleep to REPLICATE https://review.opendev.org/c/openstack/swift/+/934675 | 23:57 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!