With a granularity of 1 second, the load will from a human
perspective reflect the current situation. That also means
that the maxadmin output shows "natural" steps; 1s, 1m and 1h.
By definition, the load is calculated using the following formula:
L = 100 * ((T - t) / T)
where T is a time period and t the time of that period that the worker
spends in epoll_wait(). So, if there is so much work that epoll_wait()
always returns immediately, then the load is 100 and if the thread
spends the entire period in epoll_wait(), then the load is 0.
The basic idea is that the timeout given to epoll_wait() is adjusted
so that epoll_wait() will always return roughly at 10 seconds interval.
By making a note of when we are about to enter epoll_wait() and when we
return from it, we have all the information we need for calculating the
load.
Due to the nature of things, we will not be able to calculate the load
at exact 10-second boundaries, but it will be pretty close. And the load
is always calculated using the true length of the period.
We will then calculate 1 minute load by averaging the load value for 6
consecutive 10-second periods and the 1 hour load by averaging the load
value of 60 consecutive 1 minute loads.
So, while the 10-second load represents the load of the most recently
measured 10-second period (and not the load of the most recent 10
seconds), the 1 minute load and the 1 hour load represents the load of
the most recent minute and hour respectively.
When the parsing of a query failed, the message would treat the parameter
as a string as the printf format was `%*s` instead of `%.*s`.
The manpage of printf states the following about the precision specifier:
... or the maximum number of characters to be printed from a string
for `s` and `S` conversions.
This means that the field length specifier is somewhat meaningless for
strings.
The output of `show threads` could have a negative historic thread load
average that could be explained by the overflow of the signed 32-bit
integer used to count the number of samples.
The time that each thread started to process an event for a DCB used an
old value that is no longer used. Updating this to DCB::last_read retains
the 2.0 behavior.
According to MXS-1644 the refreshing of users may cause a crash.
In that particular case, the refreshing is triggered by the
authentication failing due to the user having hit the
'max_user_connection' limit of the server.
That is, refreshing the users in the situation is never going to
change the end-result. With this change, the users will not be
refreshed in that case and hopefully the crash will be avoided.
Note that this is something of a workaround as the crash could not
be repeated and the refreshing of the users should obviously not
ever cause MaxScale to crash.
Listed changes to MaxCtrl and the CDC connnector to release notes. Added a
note into the upgrade document about the beta versions of the MaxCtrl
client.
Placed the CDC Connector documentation in the main documentation directory
under the Connector directory. Added a link into the
Documentation-Contents.md document.
MaxCtrl now supports explicit paths for certificates and optional server
certificate verification. This allows testing by using a self-signed
certificate with the server certificate verification turned off.
Instead of the current master rejoining to the diverged master, the
current master should remain as the master server. This behavior should be
explained by the extra GTID event injected by the failover process.
The insertion and subsequent read of the data into the master in
`generate_traffic_and_check` will now be done inside a transaction. This
keeps the behavior consistent with the `check` function that only inserts
one row.
The package descriptions, summaries and licenses now use a generic
mechanism. This makes it easier to add new components that only include
small parts or use a different license.
Also updated the descriptions, the license versions and package summaries
for the main packages.
Previously, if the list contained servers that were not monitored by
the monitor yet were valid servers, an error value would be returned
and the monitor failed to start.
With this update, the non-monitored servers are simply ignored when
forming the final list.
Also, added printing of the list to diagnostics.
