The fetch command is not as reliable as it should be... Chances were that
notifications were sent multiple times if stuck in background.
Let's flush only if fully connected - and hope this fixes it.
The fetch command is not as reliable as it should be... Chances were that
notifications were sent multiple times if stuck in background.
Let's flush only if fully connected - and hope this fixes it.
For RouterOS 6.x a separate package 'ntp' exists. This adds server
functionality, but allows ip addresses for the client only. I added the
script 'rotate-ntp' to update addresses from names...
Now with RouterOS 7.x there's no extra package and the limitation does
no longer exist. So let's just drop the script.
This adds migration code, that...
* removes the script from configuration
* removes a scheduler from configuration
* sets the configured ntp pool name for ntp client
This looked smart, but did not work... 🤪
Unicode characters have to be in place to make $QuotedPrintable have an
effect. So fix it... We have duplicate pattern if symbols are disabled,
but that does not hurt.
... and in one go to limit the overall runtime.
Looks like IPv6 addresses are not flushed if the link down does not last
long enough (~ 2 seconds on linux). This results on stale addresses
after switching the vlan, which breaks connectivity.
... and in one go to limit the overall runtime.
Looks like IPv6 addresses are not flushed if the link down does not last
long enough (~ 2 seconds on linux). This results on stale addresses
after switching bridge, which breaks connectivity.
The local system time is used in the mail header. We do not want the
mails to be sent in the past, so return early (and thus wait for time
being synced).
Let's forward logs as early as possible.
All notification functions are expected to handle notifications with
queues, so nothing is lost. Just the opposite: Logs being rotated before
forwarding becomes less likely.
