Bug 2350124

Summary:	cephadm trys to bind RGW daemon to all (::) interfaces when valid networks list is provided.
Product:	[Red Hat Storage] Red Hat Ceph Storage	Reporter:	Francesco Pantano <fpantano>
Component:	Cephadm	Assignee:	Adam King <adking>
Status:	CLOSED DUPLICATE	QA Contact:	Mohit Bisht <mobisht>
Severity:	high	Docs Contact:
Priority:	high
Version:	5.3	CC:	adking, bmclaren, cephqe-warriors, c-harsh.kotak, dhill, fpantano, gfidente, jelle.hoylaerts.ext, johfulto, lonavarr, mcaldeir, mdolezel, mflusche, mkrcmari, mobisht, sostapov, trchakra
Target Milestone:	---
Target Release:	6.2
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:	2246440	Environment:
Last Closed:	2025-04-01 11:43:27 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	2233659, 2246440
Bug Blocks:	1997638, 2160009, 2236231, 2254553

Description Francesco Pantano 2025-03-05 16:47:07 UTC

+++ This bug was initially created as a clone of Bug #2246440 +++


The main issue is a problematic cephadm code path [1] where we try to verify all the IP addresses associated with a port during the RGW reconfiguration, while it should only look for specific networks.


[1] https://github.com/ceph/ceph/blob/v16.2.15/src/cephadm/cephadm#L1304

Comment 3 John Fulton 2025-03-14 13:06:28 UTC

Regarding the alertmanager issue, it seems that it is the same as with RGW. The spec contains the network list, but when alertmanager first comes up it looks like it is doing a check on *:9093. If it passes, then it binds to the network in the list.  But it will fail if opentack's haproxy is running on that port on a different network due to the initial check. The workaround is to stop openstack's haproxy to allow alertmanager to make its check and come up bound on the network, then start haproxy again.

Comment 4 John Fulton 2025-04-01 11:41:28 UTC

Regarding alertmanager: see https://bugzilla.redhat.com/show_bug.cgi?id=2274719 and its clones
Regarding RGW: see https://bugzilla.redhat.com/show_bug.cgi?id=2356355 and its clones

Lou had written:
> "something checking first for availability on 0.0.0.0:8080 as RGW is coming up before it actually binds to the network(s)"

We hope to address this in RHCSv5 via https://bugzilla.redhat.com/show_bug.cgi?id=2356354

Comment 5 John Fulton 2025-04-01 11:43:27 UTC

Adam King suggested we close this bug as a duplicate 2356355

*** This bug has been marked as a duplicate of bug 2356355 ***