From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68]) by lore.proxmox.com (Postfix) with ESMTPS id 52FD51FF136 for ; Mon, 23 Mar 2026 14:04:48 +0100 (CET) Received: from firstgate.proxmox.com (localhost [127.0.0.1]) by firstgate.proxmox.com (Proxmox) with ESMTP id 2353E18A35; Mon, 23 Mar 2026 14:05:05 +0100 (CET) Content-Type: text/plain; charset=UTF-8 Date: Mon, 23 Mar 2026 14:05:00 +0100 Message-Id: Subject: Re: [PATCH ha-manager v2 0/4] fix #2751: implement disarm/arm HA for safer cluster maintenance From: "Dominik Rusovac" To: "Thomas Lamprecht" , Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-Mailer: aerc 0.20.0 References: <20260321234350.2158438-1-t.lamprecht@proxmox.com> In-Reply-To: <20260321234350.2158438-1-t.lamprecht@proxmox.com> X-Bm-Milter-Handled: 55990f41-d878-4baa-be0a-ee34c49e34d2 X-Bm-Transport-Timestamp: 1774271054810 X-SPAM-LEVEL: Spam detection results: 0 AWL 0.347 Adjusted score from AWL reputation of From: address BAYES_00 -1.9 Bayes spam probability is 0 to 1% DMARC_MISSING 0.1 Missing DMARC policy KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record Message-ID-Hash: DXBHJ4PP7JL2EYLFBZTQBKWM3T2HRUPS X-Message-ID-Hash: DXBHJ4PP7JL2EYLFBZTQBKWM3T2HRUPS X-MailFrom: d.rusovac@proxmox.com X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; loop; banned-address; emergency; member-moderation; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header X-Mailman-Version: 3.3.10 Precedence: list List-Id: Proxmox VE development discussion List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: Thx for coming up with a v2 so quickly!=20 On Sun Mar 22, 2026 at 12:42 AM CET, Thomas Lamprecht wrote: > The biggest change compared to v1 is how ignore mode handles the service > status: instead of clearing it entirely, the relevant parts of service > status are now preserved across the disarm/arm cycle. This allows > runtime state like maintenance_node to survive, so services correctly > migrate back to their original node after maintenance ends, even if the > disarm happened while maintenance was active. Thanks @Dominik R. for > noticing this. Alongside repeating the tests conducted for v1, which again were OK,=20 I also conducted tests resembling test-disarm-maintenance{1,2,3} in a real cluster, which were also OK. > > To keep the preserved state clean, stale runtime data (failed_nodes, > cmd, target, ...) is pruned from service entries on disarm - both in > freeze and ignore mode - so the state machine starts fresh on re-arm. > The status API overrides the displayed service state to 'ignore' during > disarm-ignore mode, while the internal state stays untouched for > seamless resume. > > On arm-ha from ignore mode, the CRM now rechecks the previous resource's > node against the resource service config, picking up any manual > migrations the admin performed while HA tracking was suspended. > > First patch 1/4 is new and adds a manual-migrate simulator command as a > preparatory patch, since it is independently useful for testing the > per-service 'ignored' state handling. > > Previous discussion and v1: > https://lore.proxmox.com/pve-devel/20260309220128.973793-1-t.lamprecht@pr= oxmox.com/ > [snip] lgtm, consider this series. Reviewed-by: Dominik Rusovac Tested-by: Dominik Rusovac