Hi Sven,

Thanks for sharing your experience. Our setup is fairly small and I always 
update every piece of the cluster (acs or ceph) when I start the upgrade 
process. All acs related services are running 4.18.1.0 and all ceph services 
are on 17.2.7

Cheers

----- Original Message -----
> From: "Sven Barczyk" <s.barc...@bringe.com.INVALID>
> To: "users" <users@cloudstack.apache.org>
> Sent: Sunday, 17 December, 2023 15:13:54
> Subject: AW: KVM + Ceph volume attach problem

> Hi Guys,
> not an solution but an maybe an hint.
> 
> We made an strange experience with Ceph and ACS, after some time of usage ACS
> became slower on starting and stopping Instances, took longer than usual.
> Everthing involving attaching and detaching a Compute-Node with Ceph became
> chewy.
> Our assumption was, that after some time and with more VMs running, ACS become
> just slower.
> 
> The thing is, i started an update-spree on the OSDs without finishing it,
> because the hiccups were to bold and some customers went crazy over those
> hiccups.
> i never spend a the smallest thought about the different versions in our Ceph,
> because of its downwards compatibility.
> 
> But after some "lets update our Ceph to a newer Version", with all OSDs.
> The Problem fixed itself after the last OSD was updated. With all OSDs  and 
> Mons
> on the same Version the problem with slow ACS and ceph was fixed.
> 
> My Advice ; Keep your OSD and Mon verions clean.
> 
> KR
> Sven
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: Andrei Mikhailovsky <and...@arhont.com.INVALID>
> Gesendet: Sonntag, 17. Dezember 2023 13:45
> An: users <users@cloudstack.apache.org>
> Betreff: Re: KVM + Ceph volume attach problem
> 
> Hi Stephan,
> 
> I've checked the browser cache and it's not what is causing the problem. Done 
> a
> few private sessions and the issue persists. I've done some more testing and I
> think it might relate to some sort of timeout / delay when the web gui
> retrieves the list of vms when I try to attach a volume.
> 
> So, I click on the attach icon and I have a new modal window with a spinning
> circle. The circle spins for about 5 seconds or so and goes away. I click on
> the VM ID and it shows me about 5-6 vms. I am expecting at least about 15 in
> the list. Now, when I type the name of the vm (which is not in the list that 
> is
> shown up to begin with) in the field and wait about 15-20 seconds, the vm all
> of a sudden is shown. After that I delete the name that I've just typed and 
> now
> the list is fully populated with about 15 or so vms.
> 
> This happens every time. I've restarted the ACS, the database service, I've 
> even
> restarted the ACS server itself and the DB servers. the behaviour is
> repeatable. It happens in Safari and Chrome.
> 
> Very very strange.
> 
> Andrei
> 
> ----- Original Message -----
>> From: "Andrei Mikhailovsky" <and...@arhont.com.INVALID>
>> To: "users" <users@cloudstack.apache.org>
>> Sent: Friday, 15 December, 2023 12:44:58
>> Subject: Re: KVM + Ceph volume attach problem
> 
>> Stephan, thanks for the quick reply. I doubt the browser cache
>> cleaning was done. Let me check using Private window to see if the
>> problem is with the caches.
>> 
>> Cheers
>> 
>> ----- Original Message -----
>>> From: "Stephan Bienek" <stephan....@bienek.org>
>>> To: "users" <users@cloudstack.apache.org>
>>> Sent: Friday, 15 December, 2023 11:14:25
>>> Subject: Re: KVM + Ceph volume attach problem
>> 
>>> Hello Andrei,
>>> 
>>> we are using Ceph with RBD images on CloudStack 4.18.1.0 as well but
>>> i can not reproduce the issue you are observing.
>>> 
>>> Using the WebUI i created a new Volume, selected "Attach Disk" and
>>> all VMs are shown, running and stopped.
>>> Attaching to a running VM within the WebUI works as expected in my setup.
>>> 
>>> Probably a simple thing, but did you/the users try cleaning your
>>> browser cache after the CloudStack Update?
>>> 
>>> Best regards,
>>> Stephan
>>> 
>>>> Andrei Mikhailovsky <and...@arhont.com.invalid> hat am 15.12.2023
>>>> 11:49 CET
>>>> geschrieben:
>>>> 
>>>>  
>>>> Hello guys,
>>>> 
>>>> Any updates or thoughts on what's causing the issue with attaching
>>>> volumes in the latest ACS? I still can't attach any volumes to a
>>>> running vis from the ACS web gui. This is a bit of a problem. As
>>>> I've mentioned earlier, no errors in the management.log. The
>>>> attaching works okay from the cmk. However, most of the users that we have 
>>>> are
>>>> using web gui to manage their vms.
>>>> 
>>>> Any help would be appreciated.
>>>> 
>>>> Cheers
>>>> 
>>>> Andrei
>>>> 
>>>> ----- Original Message -----
>>>> > From: "Andrei Mikhailovsky" <and...@arhont.com.INVALID>
>>>> > To: "users" <users@cloudstack.apache.org>
>>>> > Sent: Monday, 11 December, 2023 11:39:27
>>>> > Subject: Re: KVM + Ceph volume attach problem
>>>> 
>>>> > Hey Wei,
>>>> > 
>>>> > I tested and I can attach it from the cloudmonkey just fine.
>>>> > 
>>>> > Do you know what could be causing the issue in the GUI?
>>>> > 
>>>> > Cheers
>>>> > 
>>>> > Andrei
>>>> > 
>>>> > ----- Original Message -----
>>>> >> From: "Wei ZHOU" <ustcweiz...@gmail.com>
>>>> >> To: "users" <users@cloudstack.apache.org>
>>>> >> Sent: Monday, 11 December, 2023 10:15:20
>>>> >> Subject: Re: KVM + Ceph volume attach problem
>>>> > 
>>>> >> Hi,
>>>> >> 
>>>> >> Do the volumes/vms belong to the same account ?
>>>> >> 
>>>> >> Can you try with cmk/cloudmonkey ?
>>>> >> 
>>>> >> -Wei
>>>> >> 
>>>> >> On Mon, 11 Dec 2023 at 11:05, Andrei Mikhailovsky
>>>> >> <and...@arhont.com.invalid>
>>>> >> wrote:
>>>> >> 
>>>> >>> Hello guys,
>>>> >>>
>>>> >>> I am having a strange issue which I've not noticed before. I am
>>>> >>> running ACS version 4.18.1.0 with ceph using rbd for images.
>>>> >>> I've created a new volume and trying to attach it. However, the
>>>> >>> list of VMs that I can attach a volume to is only showing
>>>> >>> STOPPED VMs. I can't seem to attach any volumes to a running vm.
>>>> >>> I've done the usual, management server logs checking, restarting
>>>> >>> the acs management server, etc. It didn't help. I've also noticed that 
>>>> >>> the nfs
>>>> >>> storage pool volumes are experiencing the same issue.
>>>> >>>
>>>> >>> I am sure that I was able to attach volumes to a running vm in the 
>>>> >>> past.
>>>> >>> What could be causing the issue?
>>>> >>>
>>>> >>> Cheers
>>>> >>>
> > > > > >> Andrei

Reply via email to