XCP-ng 8.1 Release Candidate now available!
-
The homepage welcome-screen on a 8.1.0 server point to download of XCP-ng Center 8.0.1.26.
That xcp-ng-center version cannot connect to a 8.1 server. -
@mortenchristensn I'm not sure that your test for the file type is valid. XCP-ng first creates a LV on the device then creates the ext4 filesystem in that. I assume the ext3 filesystem you found is that of the dom0 root, not that of the SR.
Here's how you can check that the local SR is ext4:
[12:36 xcp-ng-fsspniqy ~]# mount | grep XSLocalEXT /dev/mapper/XSLocalEXT--963b005a--a685--4a5f--31c0--af9a0bd98748-963b005a--a685--4a5f--31c0--af9a0bd98748 on /run/sr-mount/963b005a-a685-4a5f-31c0-af9a0bd98748 type ext4 (rw,relatime) [12:37 xcp-ng-fsspniqy ~]# file -sL /dev/mapper/XSLocalEXT--963b005a--a685--4a5f--31c0--af9a0bd98748-963b005a--a685--4a5f--31c0--af9a0bd98748 /dev/mapper/XSLocalEXT--963b005a--a685--4a5f--31c0--af9a0bd98748-963b005a--a685--4a5f--31c0--af9a0bd98748: Linux rev 1.0 ext4 filesystem data, UUID=0a1c1c21-afc4-4187-844f-7554f3f95953 (needs journal recovery) (extents) (64bit) (large files) (huge files)
-
@stevewest15 said in XCP-ng 8.1 Release Candidate now available!:
Hi,
Is there any harm to leaving the former experimental EXT4 SR driver installed for anyone who has it installed? I understand and agree with sr-create command for ext4 will fail for new SR. I think this might be a better solution for us who used the experimental ext4 and don't want to move, destroy and copy back all VMs, etc.
We leave it for 8.1 in the
sm-additional-drivers
package. But I want to remove it for future releases. It adds maintenance overhead to maintain it. If we had to maintain forever any experiment that we do then we would have to stop experimenting. That storage driver was always advertised as experimental. I wanted to provide an easy transition and I did work on it despite it being experimental so not intended for production, but it's really not simple to do and would be risky for your data. -
@mortenchristensn said in XCP-ng 8.1 Release Candidate now available!:
The homepage welcome-screen on a 8.1.0 server point to download of XCP-ng Center 8.0.1.26.
That xcp-ng-center version cannot connect to a 8.1 server.Thanks. I think that's because there's no officially stable release of XCP-ng Center out yet due do the fact that @borzel is alone to maintain is and doesn't have time for it. Either there will be a stable release of the center before our final release, or we'll update the welcome page to point to the latest dev release (with a warning).
-
We can also link the GitHub repo directly
-
@olivierlambert I would prefer that
-
"Existing SRs are untouched (so remain formatted to EXT3)" How to convert existing Local storage? I upgraded to 8.1, but Local storage is Ext3.
-
@dariosplit said in XCP-ng 8.1 Release Candidate now available!:
"Existing SRs are untouched (so remain formatted to EXT3)" How to convert existing Local storage? I upgraded to 8.1, but Local storage is Ext3.
Quoting further:
There is no easy way to convert an existing SR created with our driver, so those using it will need to move the VDIs out (to another SR or to export them), destroy the SR and create an EXT SR instead. Make sure to do this on XCP-ng 8.1.
So is your question about how to create a SR and how to destroy one? What about @Biggen's link shared above?
-
@stormi said in :
I'm not sure that your test for the file type is valid.
You are right. Running your test on my server gives same answer
Linux rev 1.0 ext4 filesystem data -
-
@mortenchristensn Local storage is an SR.
The commands below will remove the SR named "Local storage" and create a new one with the same name on the same device
If you have a pool with several hosts you need to run these commands on each host and, as stormi points out in the comment below, you need to move any VMs out of the Local SR before removing it.
# Get the IDs and device path to the current Local SR : # xe host-list | grep -B1 $HOSTNAME | grep ^uuid| cut -d\: -f2 | cut -d" " -f2 > HostUuid.txt # xe pbd-list sr-name-label=Local\ storage host-uuid=`cat HostUuid.txt` | grep sr-uuid | cut -d\: -f2 | cut -d" " -f2 > SR-Uuid.txt # xe pbd-list sr-name-label=Local\ storage host-uuid=`cat HostUuid.txt` | grep ^uuid | cut -d\: -f2 | cut -d" " -f2 > Uuid.txt # xe pbd-list sr-name-label=Local\ storage host-uuid=`cat HostUuid.txt` | grep device-config| cut -d\: -f3 | cut -d" " -f2 > Device.txt # Unplug and destroy the Local SR : # xe pbd-unplug uuid=`cat Uuid.txt | egrep "*-*-*-*-*" ` # xe sr-forget uuid=`cat SR-Uuid.txt | egrep "*-*-*-*-*" ` # Create the new Local SR : # xe sr-create content-type=user device-config:device=`cat Device.txt` host-uuid=`cat HostUuid.txt` name-label=Local\ storage shared=false type=ext
-
Of course, move the VMs out of the SR before destroying it
-
@peder Dosen't work. My version is 8.1 Build date: 2020-02-19 and reciveed the resoult " Name: Local storage Type: Ext3."
-
@dariosplit What exactly doesn't work? Where did you receive that output?
-
@Danp I got Ext3 again instead of Ext4.
-
@dariosplit said in XCP-ng 8.1 Release Candidate now available!:
@Danp I got Ext3 again instead of Ext4.
How do you check exactly? (you haven't answered to @Danp's "Where did you receive that output?")
-
@stormi cat /etc/fstab
LABEL=root-penujp / ext3 defaults 1 1
LABEL=swap-penujp swap swap defaults 0 0
LABEL=logs-penujp /var/log ext3 defaults 0 2 -
That's
/
and/var/log
, not a local storage repository. Just read the other messages of this thread for a way to find the local SR and test its FS type. -
After an ISO upgrade to RC1 from 8.0 it booted... so that's good.
Using XCP-ng Center 99.99.99.28:
Changing "Control Domain Memory" does not take effect.
SMB ISO fails "SM has thrown a generic python exception" for existing SR.
I deleted the ISO SR and can't create a new one (same error):
SMB ISO fails "SM has thrown a generic python exception" for new SR.Python errors at boot:
Traceback (most recent call last):
File "/usr/lib64/xsconsole/XSConsole.py", line 46, in <module> main()
File "/usr/lib64/xsconsole/XSConsole.py", line 37, in main app.Enter()
File "/usr/lib64/xsconsole/XSConsoleTerm.py", line 57, in Enter Data.Inst(),Up
date()
File "/usr/lib64/xsconsole/XSConsoleData.py", line 343, in Update self.DeriveDa
ta()
File "/usr/lib64/xsconsole/XSConsoleData.py", line 361, in DeriveData name = "
".join(cpu['modelname'].split()))
TypeError: string indices must be integers, not strI did not have the problems with 8.0.
I had one VM that will no longer boot because of a missing disk. After attaching the disk there was a message about a PV error (it's not set for PV) and it still fails.
As a test system there may have been some issues left over but it was all working with 8.0.
So as a "production ready" upgrade. No.
-
Thanks for the feedback.
What's the output of
uname -a
?@Andrew said in XCP-ng 8.1 Release Candidate now available!:
After an ISO upgrade to RC1 from 8.0 it booted... so that's good.
Using XCP-ng Center 99.99.99.28:
Changing "Control Domain Memory" does not take effect.We do not support XCP-ng Center. The community as a whole can. Can you reproduce with other methods of changing dom0 memory? See https://support.citrix.com/article/CTX134951
SMB ISO fails "SM has thrown a generic python exception" for existing SR.
I deleted the ISO SR and can't create a new one (same error):
SMB ISO fails "SM has thrown a generic python exception" for new SR.Does it work from Xen Orchestra? From command line using
xe
? Can you extract the errors at the time of the action from/var/log/SMlog
,/var/log/daemon.log
and/var/log/xensource.log
?Python errors at boot:
Traceback (most recent call last):
File "/usr/lib64/xsconsole/XSConsole.py", line 46, in <module> main()
File "/usr/lib64/xsconsole/XSConsole.py", line 37, in main app.Enter()
File "/usr/lib64/xsconsole/XSConsoleTerm.py", line 57, in Enter Data.Inst(),Up
date()
File "/usr/lib64/xsconsole/XSConsoleData.py", line 343, in Update self.DeriveDa
ta()
File "/usr/lib64/xsconsole/XSConsoleData.py", line 361, in DeriveData name = "
".join(cpu['modelname'].split()))
TypeError: string indices must be integers, not strWhere is that error seen? Is it visible directly in the console in place of XSConsole? It's like the value of
cpu
is a string instead of a dict. Weird.I did not have the problems with 8.0.
I had one VM that will no longer boot because of a missing disk. After attaching the disk there was a message about a PV error (it's not set for PV) and it still fails.
I've never seen upgrades detach disks from VMs. I suggest to open a separate thread for discussing and debugging the issue about this VM. Please give as much information as possible in that thread, including errors from logs (see https://github.com/xcp-ng/xcp/wiki/Logfiles).