| opendevreview | Tim Burke proposed openstack/swift master: ec: Refuse to sync non-durables that are fresh on disk https://review.opendev.org/c/openstack/swift/+/796982 | 01:24 |
|---|---|---|
| opendevreview | Matthew Oliver proposed openstack/swift master: memcache: Add an 80% item_size_max warning https://review.opendev.org/c/openstack/swift/+/795915 | 06:44 |
| opendevreview | Merged openstack/swift master: Add async_pending_last time to object.recon https://review.opendev.org/c/openstack/swift/+/758603 | 08:11 |
| DHE | if I run servers with a different replication port, does servers_per_port listen independently on the main port and the replication port? using a unique replication port doubles the number of worker processes? | 12:46 |
| DHE | I'm trying to find the code and am not sure.. | 12:46 |
| DHE | experimenting... the answer seems to be no... | 14:15 |
| timburke | DHE, yeah, i'm not sure those features work well together, unfortunately :-( | 15:00 |
| timburke | see also, https://bugs.launchpad.net/swift/+bug/1669579 and https://review.opendev.org/c/openstack/swift/+/337861 | 15:00 |
| DHE | disappointing... | 15:19 |
| zaitcev | You know what I think is coming down the pipe in Swift | 16:58 |
| zaitcev | By looking at all the work-division things that have processes=8 process=1 | 16:59 |
| zaitcev | We're going to end with a recon module that 1) compares if all nodes have same processes=N, 2) if the number of nodes running expirer/reconciler whatever equals N, 3) the process= of each node occurrs only once. | 17:01 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Return 503 for container listings when shards are deleted https://review.opendev.org/c/openstack/swift/+/793492 | 18:50 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Return 503 for container listings when shards are deleted https://review.opendev.org/c/openstack/swift/+/793492 | 18:51 |
| ade_lee | timburke, any idea what might be going on here -- https://21fbb9beee6499333383-b4831d08abece0047714b6befcdb357a.ssl.cf5.rackcdn.com/796057/3/check/swift-tox-func-encryption-py36-centos-8-fips/8df3976/job-output.txt ? | 18:59 |
| ade_lee | timburke, seems like the other two test jobs passed and this one errors out with "No space left on device" | 19:00 |
| timburke | looking | 19:03 |
| timburke | so we're definitely creating and mounting an XFS filesystem to be used for tests... and there's enough freespace on the root partition that we shouldn't run into trouble where we've created this sparse 1G file but can only use a few hundred MB of it | 19:13 |
| timburke | i know the encryption job is going to want to write down a lot more metadata in xattrs, and that seems to be where it's raising the ENOSPC -- i wonder if something changed with how much space xfs reserves for inodes on that tiny volume between centos 7 and 8? | 19:16 |
| ade_lee | timburke, hmm not sure | 19:19 |
| DHE | I have machines running both OSs. what command would print that? | 19:20 |
| timburke | :-/ meanwhile, some of the passing jobs say things like "test.functional.test_account.TestAccount.test_bad_metadata3 ... SKIPPED: Large xattrs not supported in `/home/zuul/xfstmp`. Skipping test" | 19:21 |
| timburke | DHE, i think comparing the outputs of something like `truncate -s 1GB some-file && mkfs.xfs some-file && rm some-file` might be useful -- though we've got that in the zuul logs: http://paste.openstack.org/show/806785/ | 19:31 |
| timburke | there's a difference, but it's not obvious to me that centos8 shouldn't work | 19:31 |
| timburke | i wonder if there's a chance FIPS-mode could interfere with xattrs? | 19:32 |
| ade_lee | timburke, we can certainly run a job on centos without fips at the same time | 19:35 |
| opendevreview | Ade Lee proposed openstack/swift master: WIP/DNM: Add FIPS CI jobs https://review.opendev.org/c/openstack/swift/+/796057 | 19:55 |
| opendevreview | Ade Lee proposed openstack/swift master: WIP/DNM: Add FIPS CI jobs https://review.opendev.org/c/openstack/swift/+/796057 | 20:05 |
| opendevreview | Clay Gerrard proposed openstack/swift master: Handle EC disconnect better https://review.opendev.org/c/openstack/swift/+/796735 | 20:14 |
| opendevreview | Merged openstack/swift master: Allow floats for a couple more intervals https://review.opendev.org/c/openstack/swift/+/794893 | 23:06 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!