The console tab kept its own `minimalLayout` state to track whether the page header was collapsed, separate from the VM page's `collapsedHeader`. The two desynced because the tab remounts on navigation while the VM page stays mounted.
Fix:
- Single source of truth `collapsedHeader` in the parent index.js
- Pass it to the child tab-console.js along with `setCollapsedHeader`
- Replace the `_toggleHeader` toggle with an explicit `_setCollapsedHeader(value)` setter
- Remove the duplicate local `minimalLayout` state that could get out of sync and use `collapsedHeader` instead
- Force off on VM shutdown (console no longer displayed) and on unmount (user changed page)
The docs on devops tools and kubernetes was moved
from the Vates VMS doc to the Xen Orchestra doc.
As a result, one link to the kubernetes page broke in the XO doc.
This commit fixes that broken link.
Signed-off-by: Thomas Moraine <thomas.moraine@vates.tech>
* add more details in task log
* fix an issue with VM with more than 10 snapshot, add unit test for this parsing
* fix the real bug: vmware do not update the vmdk description file before the next VM reboot. It means that the uid of the snapshot and vm are the same, thus finishing the transfer with " nothing to transfer" . We are now using the path on source as key + add unit test + add compatibiliy with legacy VM migrated with the uid.
* add e2e test for v2v
Customer imported large VM from VMWare (> 3 To), import switched to Qcow2 automatically but the process was stuck while building the qcow2 L1/L2 address tables, where the NbdDisk's hasBlock is called for each l2 table entry which iterated over the disk dataMap which contained ~113,528 ranges. This froze the process for more than 18 minutes after which it was either killed or died due to OOM error.
This fix improves the NbdDisk by:
Updating processDatamap to throw on overlapping ranges and merge touching extents.
Override default getBlockIndexesCount with optimized Nbd one.
Improve hasBlock with a built in cursor to improve performance.
Test file added to cover the new function and updated code.
from ticket 59505
node-tar >= 7 returns a Minipass stream: it has `pipe()` but is not a `node:stream`
instance. xo-proxy serves this stream through Koa, which pipes only bodies passing
instanceof Stream` and JSON-serializes anything else — so a file restore through a
proxy answered a JSON dump of the tar object instead of the archive. Wrapping it in a
real node stream also propagates tar errors to the consumer and aborts the tar job
when the client disconnects.
The load balancer did not take into account that hosts may be in disabled state, which means it's impossible to migrate a VM to it.
This PR fixes that by ignoring these hosts instead.
Fixes ticket 62077
* feat(xo-server): serve doc locally
* the doc is already distributed from the mono repo
* add a build phase to the global yanr build that build a doc with the
right prefix
* add a http mount doc => embed build
* doc(backups): add a space calculator
* estimate the space needed for backups
* compute the duration protected by immutability
