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 ADEEF1FF165 for ; Thu, 6 Nov 2025 12:57:11 +0100 (CET) Received: from firstgate.proxmox.com (localhost [127.0.0.1]) by firstgate.proxmox.com (Proxmox) with ESMTP id 3A83714E03; Thu, 6 Nov 2025 12:57:52 +0100 (CET) Message-ID: Date: Thu, 6 Nov 2025 12:57:17 +0100 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird To: Nicolas Frey , Proxmox Backup Server development discussion References: <20251105155129.517430-1-n.frey@proxmox.com> <20251105155129.517430-4-n.frey@proxmox.com> <685be513-ba3e-4c78-9ed1-8ff7861aaad0@proxmox.com> <8f047692-43b3-4ec0-8b19-17729172d860@proxmox.com> <48b1a0a9-314f-4d4d-b2ed-1e3444d6dced@proxmox.com> Content-Language: en-US, de-DE From: Christian Ebner In-Reply-To: <48b1a0a9-314f-4d4d-b2ed-1e3444d6dced@proxmox.com> X-Bm-Milter-Handled: 55990f41-d878-4baa-be0a-ee34c49e34d2 X-Bm-Transport-Timestamp: 1762430219042 X-SPAM-LEVEL: Spam detection results: 0 AWL 0.046 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 RCVD_IN_VALIDITY_CERTIFIED_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to Validity was blocked. See https://knowledge.validity.com/hc/en-us/articles/20961730681243 for more information. RCVD_IN_VALIDITY_RPBL_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to Validity was blocked. See https://knowledge.validity.com/hc/en-us/articles/20961730681243 for more information. RCVD_IN_VALIDITY_SAFE_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to Validity was blocked. See https://knowledge.validity.com/hc/en-us/articles/20961730681243 for more information. SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record Subject: Re: [pbs-devel] [PATCH proxmox-backup 2/4] api: verify: use worker-threads to determine the number of threads to use X-BeenThere: pbs-devel@lists.proxmox.com X-Mailman-Version: 2.1.29 Precedence: list List-Id: Proxmox Backup Server development discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: Proxmox Backup Server development discussion Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii"; Format="flowed" Errors-To: pbs-devel-bounces@lists.proxmox.com Sender: "pbs-devel" On 11/6/25 12:21 PM, Nicolas Frey wrote: > On 11/6/25 10:32 AM, Christian Ebner wrote: >> On 11/6/25 10:22 AM, Nicolas Frey wrote: >>> On 11/6/25 10:08 AM, Christian Ebner wrote: >>>> Please add a short commit message describing what the worker threads >>>> cover, e.g. that this parameter controls the number of reader and >>>> chunk verification threads. >>>> >>>> What tripped me over just now: >>>> Is this intentionally not increasing the number of chunk verification >>>> threads? Or was that overlooked? From the name of the parameter I >>>> suspected this to act on both, reading and verifying. If this is not >>>> the case, maybe the parameter should get renamed to a more telling >>>> `parallel-chunk-readers` instead? >>> >>> I wasn't sure if the number of threads for verification should be >>> controlled via this as well, as the original patch only added a new >>> thread pool for reading, whereas the verification pool was already >>> implemented. >>> I pointed this out in the cover letter, though it might have been >>> better to put this here too: >>> >>> The number of `worker-threads` only controls the thread pool for >>> reading, though if it makes sense to reuse this for the verification >>> pool, it could be adjusted to do so too. >>> >>> I think it makes sense to use it to control the number of threads of >>> both. Thanks for the feedback, I'll adjust it along with the other >>> proposed changes in a v2! >> >> Well, that was just an uninformed assumption from my side when reading >> the parameter name (and I did not re-read the cover letter today after >> having looked at this quickly yesterday, sorry for that). > > That makes sense, the parameter name does not accurately describe the > function it serves here anyway, so that should have been named a bit > better. > >> >> But maybe you can also evaluate if it actually makes sense to control >> both by the same parameter, or if it only makes sense to e.g. increase >> the number of verification tasks (no point for that if the IO remains >> the bottleneck), or if it would make sense to have either 2 parameters >> or couple them by some proportionality constant. >> > > I had an idea along the lines of: > > self.worker_threads.mul(2).clamp(4, 32), On second thought, this will most likely not cover most cases? One system could be severely IO bound, the other one severely CPU bound... > though the proportionality factor should be tested to determine what > would actually be sensible here and of course be documented accordingly. > > I also thought a minimum of 4 threads for verification makes sense, as > when the default value of 1 thread is used, it has somewhat the same > behavior as before adding the read thread pool (i.e. 1 thread for > reading, 4 threads for verification) and would scale somewhat > accordingly. The threads should also clamped to a max of 32 to respect > the constraints of the schema also stating 32 as a max. > > What do you think? I think it would make sense to keep both decoupled for the time being, especially since this might depend strongly on the backend. E.g. for S3 backed datastores you might gain a lot by increasing the number of readers, but not much by increasing the number of verify threads. _______________________________________________ pbs-devel mailing list pbs-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel