@Shrubbery said in xcp-ng 8.3 not recognizing all host memory:

The issue was memory reserved for the integrated graphics card

haha.. that is such a silly thing... but good to know then next time I'm working on some hardware and wondering where the RAM is going too...

@flakpyro said in Manual snapshots retention:

@olivierlambert I wonder if it would be beneficial to show all snapshots older than 30 days not just ones not created by an automated process. For example what happens if a XO backup job runs, creates a snapshot but the process is interrupted and the job fails. Will the next time the job runs clean up the previous failed jobs snapshot or is there a chance a snapshot could be left behind?

Garbage collection should be handling this...

@olivierlambert I added a response in the hopes that it clarifies the end-goal.

@Tristis-Oris said in XO tasks - select all button:

@olivierlambert i remember news, but it never works for me(

Are you using XOA or XO from Source?

@manilx said in Manual snapshots retention:

@Tristis-Oris To be honest, this just opens more "configurations/complications" for a fringe case.....
Manual backups should never be deleted, unless I specifically do that (in an enterprise environment you don't necessarily know why they were made but there certainly was a reason fir them).

An information in the dashboard for backups "older" than x days could be interesting, yes.

Right, I think having just a list of Snapshots older then X (that aren't created from a scheduled job) would be beneficial to validate that things aren't getting too long in the tooth.

@CodeMercenary Do you mind sending me your serial number, I'd be happy to take a look for you to confirm.

If you don't want any jobs running in parallel at all, you would set your concurrency to 1 (or maybe 0) so that only a singular VM would be completely backed up at a given time, but this will certainly break the "Start my Backups at 01:00".

Because your backup jobs are starting at 01:00 for the first VM chosen, but your last VM might not start it's backup until 23:00 the following day depending on your system and network performance...

@cgasser here is an article on it that @olivierlambert wrote some time ago, https://xen-orchestra.com/blog/xen-orchestra-backup-concurrency/

Basically you have to set your concurrency to be lower than the default limit of 2 per the documentation. https://docs.xen-orchestra.com/backups#backup-concurrency

@cgasser I believe this was discussed in another thread, but to summarize you should set your job-concurrency, otherwise it defaults to something like 4 jobs at once.

But even in that scenario, not every job is going to run at 01:00 because some job would be waiting until the last job has completed, offsetting your start time of the next job.

@Shrubbery If you check BIOS do you see the full 16GB of ram, and does this box have a built in memtest that you can validate that the memory isn't faulty?

Wouldn't this be incredibly dangerous?

While I wouldn't be opposed to seeing a list of all VM's with manually created snapshots, I wouldn't want the system automatically deleting a snapshot that I made, because I might have need that snapshot for forensic and investigation purposes or some other legal requirement.

Not that I have that requirement today, but it's plausible that as an organization it could happen.

@CodeMercenary said in Seeking advice on debugging unexplained change in server fan speed:

@DustinB Nothing useful yet. I rebooted the servers and explored a bit in the BIOS to see if there were any settings, or to at least tweak some things to see if it would reset whatever went wrong in the reboot in mid December. While doing that I found that one of the two impacted servers was a version behind for the BIOS as well as for the iDRAC so I updated both of them. Unfortunately, that made no change to the fan speeds.

I've been out sick all of this week, so far, but I'll be looking into this more when I get back to the office. I've read about ways to manually control the fans but I'd rather not have to depend on a script running somewhere that makes those kinds of decisions, I'd much rather have iDRAC, or whatever normally controls it, handle it like it used to.

Sorry you're not feeling well, when you're back on your feet, specifically look for Firmware for your Fans.

Hope you're feeling better soon.

Passing a PCI GPU or a USB device to a given VM basically forces this, since these cards are Host Specific and are unable to be shared across hosts in the same pool.

IIRC that is..

@irtaza9 You can opt to have a VM have a preferred Host Affinity, which basically means "Run here if possible" using the advanced settings while creating a VM or the Advanced settings once the VM has been created.

@manilx said in Price Increases:

@olivierlambert said in Price Increases:

Comparing us to VMware feels unfair given our dedication to an open-source approach, a stark contrast to closed, proprietary systems. We’ve always strived to support smaller organizations while advancing open-source innovation in virtualization.

Actually it's more than just unfair

Only because VMWare's lowest tier so much worse than any edition of XO with XCP-ng, and it (VMware) doesn't (or didn't include any support) and that all pricing models going forwards are way more expensive for the bare basic hypervisors. If you require some additional feature of ESXi at different tiers, you wouldn't even be considering any other hypervisor platforms, because ESXi does this specific thing and nothing else does.

Even using XCP-ng and XO(CE) with community support gets you more for literally free.

So yeah, I guess it's really unfair, free and open source with the option for paid support, or Forced subscription at the lowest tier and no official support.....

@irtaza9 said in Hosts vs. Pools - Information Required:

xcp-ng is a distribution which runs on a bare-metal and allow us to perform virtualization. I want to know more about hosts, pools. What is the difference between these two concepts. Also what is cluster and What about HA? Where does the HA happen? b/w pools or b/w hosts?

A single host, is a pool of 1 host. A pool can contain multiple hosts so that VMs can be live migrated from host to host (best performance with shared storage for your VMs).

HA can be achieved through a variety of options, using XOStore from the Vates team for a licensing fee. Is one such solution, but this is Sharing your Storage.

HA doesn't mandate HA Storage or Shared Storage, but it definitely helps.

@ddbolt deploying and maintaining XO from source is stupidly simple, and well documented. I even have one of the if not the original deployment script that I helped to write on my github.

@ddbolt said in Price Increases:

@manilx
I certainly see your point.
I just wish there were something in middle where they keep me patched and my XO updated and maybe ai just pay for support on an “as needed” basis.
I compiled it from source one a year ago and it did work but it would just be nice to have them keep my software up to date for a smaller fee and I just pay for additional support as needed. Like $300 per host per year but I would pay additional to speak with someone at Vates.
That I could do.

Vates has said that a per-incident response isn't viable at this time several times because of the ad hoc nature and the potential un-scalability of such a model.

The community here offers a lot of help, and as others have said you can deploy XO from source and not pay a dime for it. Helping you to save all of the cost until you can get your customers to start paying for said services.

@olivierlambert I agree wholeheartedly with you on that. Keeping the system stock is best for support.

Separately, is there any planned work on officially integrating support for Uninterruptable Power Supplies and XCP-ng 8.3?

@olivierlambert While XCP-ng is it's own distribution, it looks like it was built from redhat, as I can find redhat-release in /etc and when I cat the file I get xcp-ng-8.2.1 as the output.

Maybe some cleanup remains. . .