| zaitcev | Guys | 00:56 |
|---|---|---|
| opendevreview | Merged openstack/swift master: Plumb allow_modify_pipeline through run_wsgi/run_server https://review.opendev.org/c/openstack/swift/+/804544 | 00:56 |
| zaitcev | Does out in-tree s3api support bucket-in-hostname addressing, like bucket.s3.myhost.com? | 00:57 |
| zaitcev | I was sure we did, in fact I arrangements made in DNS with the wildcard. | 00:57 |
| zaitcev | But I cannot find the place where we look at Host: or env['HTTP_HOST'] or requiest.host that is relevant to this. | 00:58 |
| zaitcev | actually | 00:59 |
| zaitcev | self.bucket_in_host = self._parse_host() | 00:59 |
| zaitcev | Spoke too soon. | 00:59 |
| opendevreview | Tim Burke proposed openstack/swift master: s3api: fix the copy of non-ASCII objects https://review.opendev.org/c/openstack/swift/+/811388 | 02:05 |
| opendevreview | Merged openstack/swift stable/xena: Update .gitreview for stable/xena https://review.opendev.org/c/openstack/swift/+/810825 | 02:51 |
| opendevreview | Merged openstack/swift stable/xena: Update TOX_CONSTRAINTS_FILE for stable/xena https://review.opendev.org/c/openstack/swift/+/810826 | 03:31 |
| opendevreview | Matthew Oliver proposed openstack/swift master: container-updater: demonstrate account stat blockage with probe test https://review.opendev.org/c/openstack/swift/+/811833 | 03:53 |
| opendevreview | Pete Zaitcev proposed openstack/swift master: Get rid of port to node assumptions and their modulo kludges https://review.opendev.org/c/openstack/swift/+/811876 | 05:42 |
| opendevreview | Merged openstack/swift master: s3api: fix the copy of non-ASCII objects https://review.opendev.org/c/openstack/swift/+/811388 | 05:58 |
| *** elodilles_pto is now known as elodilles | 07:12 | |
| reid_g | We are looking to handle these timeouts: ERROR with Object server 6000/d0 re: Trying to get commit status of PUT to /v1/A/C/O: Timeout (10.0s). Would the proxy_server node_timeout be the correct one? | 15:56 |
| reid_g | recoverable_node_timeout - is that incase a node doesn't respond (fast enough) then the proxy will try to connect to another node to perform the GET/HEAD request? | 16:58 |
| timburke | reid_g, yeah, you want node_timeout for those commit status timeouts. and yeah, recoverable_node_timeout is about issues mid-stream during GET | 18:02 |
| timburke | (for initial responses for GET/HEAD, i'd look more at concurrency_timeout and concurrent_gets) | 18:03 |
| reid_g | mid-stream - meaning GET issued -> Proxy connects to nodes to get frags -> a node times out -> try another? | 18:57 |
| reid_g | Also are there logs we should look for to know if we need to enable concurrency? | 18:57 |
| reid_g | Was just checking configurations and it appears we already have concurrent_gets = True with the default timeout | 19:32 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!