| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
int is always the native word size, and we don't really test anything on
32bit OSes anymore. 32bit support already required ugly workarounds in
the past for large file sizes and ip2long, so we can finally get rid of
those and just put an initial check in install.php.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
When a client sends its current VNC password, we look up the client's
uuid via the remote IP address. This would return any random client that
last had this IP address when it booted. This is mostly not a problem
with clients that have a static assignment, but can easily go wrong with
clients in dynamic pools, as it's likely we have a few entries with the
same address in the DB.
We now get the client that was seen active most recently, which–unless
we have address collisions–should always be the proper one.
|
|
|
|
| |
References #3767
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
- Locations that are set to "never" or "only outside business hours"
will not be considered during WOL, and not delivered to the guacamole
proxy.
- If we cannot wake as much machines per remote group as configured,
we'll display a little number next to the WOL count setting.
|
|
|
|
|
|
|
| |
This doesn't make any sense; we already account for this by looking at
the woltime column.
This reverts commit 0c561cd07c82d09ec5f6f1aa0a92ead403d0952b.
|
|
|
|
|
|
|
|
| |
If a location is assigned to multiple groups, we potentially wake too
many machines, as the number of clients that received a WOL in that
location when processing group A is not accounted for when peocessing
arrives at location B. Keep track of the number of WOLed machines per
location to be able to avoid this situation.
|
|
|
|
|
| |
.... so the proper location information can be displayed on the WOL task
status page.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Remoteaccess mode can now be forced to be disabled for individual
locations in locations module, either unconditionally, or whenever the
openingtimes schedule says the room is open. A reboot will be triggered
whenever the room opens/closes to force clients into the proper runmode.
|
| |
|
|
|
|
|
|
|
| |
* Auto-convert to utf8mb4_unicode_520_ci
* Use foreach instead of while to loop over results
* Drop useless statement caching
* Keep emulated prepares, as we sometimes loop over nested queries
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This way after 30+ minutes we would still pop up the screen saver that
would count down to the next logout/reboot/shutdown, so the user gets
an idea that the session won't last forever.
|
|
|
|
| |
References #3823
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
Enable rebootcontrol module in ensureMachinesRunning(),
so it works from the call in api.inc.php.
Consider machines that we sent a WOL packet to during the last 90
seconds as online when counting number of idle machines, so slow
booting machines won't cause us to wake too many machines.
|
| |
|
| |
|
| |
|
|
|