This file seems to have an unmatched "td" ending. Adding the line here matches the "td" at line 320 and this embraces the little table that has the 4 icons in it in a square that comes at the right hand end of each port-forward entry in the main table.
I can't see any difference in the rendering of the page, at least on Firefox, with and without this fix.
The tabbing of this file is woeful. I am fixing that up with a code-style review. But thought I should do a separate pull request for this kind-of-functional fix.
If the DHCP IP address is in a pool (not in the main DHCP range for the interface) then the interface that corresponds to the IP address is not found. This results in the link to "add static mapping for this MAC address" not having any value for "if=" and thus clicking on the "+" button does not work.
Reported in bug 4649
Process any pools when checking for which interface contains the IP address.
If I go to Service->DHCP Server, make some edits that are invalid (e.g. change range start or end to some invalid string) and press Save then the page comes back displaying the input error(s). But it also says:
"The changes have been applied successfully."
Actually, the changes (which were invalid) have not been applied to the config - all is well there - but dhcpd has been stopped and started and dnsmasq or unbound has been kicked... which is all unnecessary processing since the user has not yet provided valid values to save.
Actually the GUI is displaying the opposite setting to what is in the config. When the user pressed save that opposite setting was saved, but then again it displays the opposite of the opposite...
I noticed this when cleaning up VIPs and OpenVPN server when testing for this forum post https://forum.pfsense.org/index.php?topic=92174.0
The system let me delete my test VIP before I deleted the OpenVPN server using, probably not a good thing.
This is a better version of #1604 - the similar loops for server and client are combined, and the error message shows the description (or IP address:port) of the offending OpenVPN instance so the user has a clue about which OpenVPN instance uses the VIP.
Tell users that this is possible in DNS Resolver and how to achieve it. The code in unbound.inc already supports it and works.
I had asked for this in Redmine feature request #4350 and when I went to look at coding to implement it I found code that already did it.
So IMHO it is worth telling users.
Note that advertise is spelt with an "s" in other places in the GUI, so
making it consistent in services_ntpd - but maybe Americans do spell it
"advertize" these days?
this helps pick a free port for services using sockets bound to localhost, and helps determine if the service has at least started and bound the port without needing to go through all 'connected' sockets as well
https://forum.pfsense.org/index.php?topic=91168.msg505273#msg505273
$config['voucher'][$cpzone]['msgnoaccess']
and
$config['voucher'][$cpzone]['msgexpired']
do not exist.
These should be
$config['voucher'][$cpzone]['descrmsgnoaccess']
and
$config['voucher'][$cpzone]['descrmsgexpired']
The other tabs of Status:RRD Graphs put the friendly description of each interface into the drop-down list for selection.
This change makes the Custom tab do that also.
Note: We can let the code pass "never" (or any other unexpected stuff)
to adjust_gmt()
adjust_gmt() should anyway handle the case when strtotime() cannot
understand the input string and thus returns false. In that case we
return the input string as-is so it will be displayed as the time. That
way the user will see it and can report easily whatever other unexpected
char data was in the leases file.
It also prevents "false" (zero) being converted to the date-time string
and thus becoming the Unix epoch 1 Jan 1970 on the display.
Latest forum report of this kind of thing:
https://forum.pfsense.org/index.php?topic=90083.0
This code just looked wrong. It was considering 10.1-RELEASE-p6 to be release number "1" and comparing it to "9".
These changes to do what it seems to intend. This will make that UFS+J stuff appear, if that is of any consequence.
1) Only attempt to delete the oldusername if it actually was non-empty - at the moment errors are logged in the system log when adding a new user, because the code was trying to delete the user name "".
2) Call local_user_set() first to create (change, whatever) the user record. This makes the user record exist for a new user. Then call local_user_set_groups() to sort out what groups the user should be in or not in. The existing code would fail to add a new user to the specified group/s because local_user_set_groups() was called too early, before the user actually existed.
Typical system log errors from the old code:
Mar 18 17:10:31 php-fpm[9542]: /system_usermanager.php: Tried to remove user but got user pw instead. Bailing.
Mar 18 17:10:31 php-fpm[9542]: /system_usermanager.php: The command '/usr/sbin/pw groupmod admins -g 1999 -M '0,2003,2006,2008' 2>&1' returned exit code '67', the output was 'pw: user `2008' does not exist'
From looking at the code history, I think this has been this way for a long time, not a new bug at all.
Discussed in forum: https://forum.pfsense.org/index.php?topic=90700.msg501766#msg501766
At the moment you can make a VLAN with tag 0. The input validation does not catch it because when $_POST['tag'] = "0" that evaluates to false by PHP.
Always make the checks on 'tag' value whenever the 'tag' key is set at all. If the (required) 'tag' key is not set, then that is already checked for by do_input_validation().
This just looks wrong. But I guess the code path never comes through here because function readline() already exists in the environment of this script.
