ID: 11093
Title: aws_cloudwatch_alarms_limits: fix discovery
Component: Checks & agents
Level: 1
Class: Bug fix
Version: 1.7.0i1
This werk fixes an issue with the discovery of the service
aws_cloudwatch_alarms_limits, which was never discovered, even if
the AWS special agent was configured accordingly. This service
monitors the number of AWS CloudWatch alarms.
ID: 11092
Title: One service per region for AWS limits
Component: Checks & agents
Level: 1
Class: New feature
Version: 1.7.0i1
With this werk, the monitoring of limits for AWS
services is split into one service per region. Before,
all regions were combined into a single service, which
resulted in potentially quite a lot of output and
caused problems with the graphs for 2 or more regions.
To apply this werk, users have to re-discover the
corresponding services, which will result in a loss
of the historically recorded metrics. This is best done
using "Automatic refresh (tabula rasa)" in the WATO-menus
of the affected hosts. Note that rules for upper or lower
levels created before this werk are not affected and will
apply to all newly discovered services. After installing
the werk, they can be modified to only apply to certain
regions.
ID: 11104
Title: heartbeat_crm: go to CRIT if data is too old
Component: Checks & agents
Level: 1
Class: Bug fix
Version: 1.7.0i1
The service previously went to UNKNOWN instead of CRIT.
ID: 10535
Title: Activate Changes: Now able to specifiy the number of parallel site activations
Component: WATO
Level: 1
Class: New feature
Version: 1.7.0i1
Earlier versions may have run into memory problems when too much sites were activated at the same time.
Each site was handled in a separate helper process. This consumed quite a lot of memory when your
WATO setup included lots (>50) of sites.
You can now configure how much helper processes are spawned during the activate changes phase.
If there are more sites to activate then available helper processes, the remaining sites are put into a
queue and will be processed once a helper process becomes idle again.
Per default the optimal number of helper processes is calculated automatically (up to 90% remaining RAM is used),
which should be OK for most setups. The new global settings option <tt>Maximum parallel processes during activate changes</tt> allows
you to adjust this value. The hardcoded minium value is 5.
ID: 11103
Title: systemd_units: Fix parsing error
Component: Checks & agents
Level: 1
Class: Bug fix
Version: 1.7.0i1
In some cases the parsing of the systemd_units plugins output crashed with an IndexError. This has been fixed.
ID: 10781
Title: Additional calendar aware time ranges
Component: WATO
Level: 1
Class: New feature
Version: 1.7.0i1
Particularly useful for Metric History Painters this date aware time ranges
have been added:
<ul>
<li>7 days back (this day last week)</li>
<li>8 days back</li>
<li>2 weeks back</li>
</ul>
ID: 11027
Title: LDAP: Validate usage of suffix
Component: WATO
Level: 1
Class: Bug fix
Version: 1.7.0i1
If you configure multiple LDAP connections with the same suffix, the 'Users'
page is broken and can not be accessed any more. From now on, you will be
warned, if a suffix is already in use by another connection.
ID: 10906
Title: Extend UCD SNMP scan function
Component: Checks & agents
Level: 1
Class: New feature
Version: 1.7.0i1
The UCD SNMP scan function got extended to detect devices without system description OID.
This change allows to detect devices like Dell EMC ECS systems.
For these special devices make sure to set the rule "Hosts without system description OID".
However this should normally be avoided if possible.
ID: 10987
Title: HW/SW Inventory Service: Fix "State when inventory fails"
Component: HW/SW Inventory
Level: 1
Class: Bug fix
Version: 1.7.0i1
The state of the HW/SW Inventory service can be determined by
{{State when inventory fails}} in the ruleset {{Do hardware/software Inventory}}.
This state was overwritten by the following error cases:
<ul>
<li>Empty output, default CRIT</li>
<li>Connection, default CRIT</li>
<li>Timeout, default CRIT</li>
<li>Exception, default UNKNOWN</li>
</ul>
This behaviour was introduced with 1.5.0 and and has been fixed now:
The HW/SW Inventory service state won't be overwritten any more.
ID: 11091
Title: AWS monitoring: adjustments to werk #10962 and bugfixes
Component: Checks & agents
Level: 1
Class: Bug fix
Version: 1.7.0i1
This werk adjusts the output several checks for AWS
services. With werk #10962, the output of these checks
changed from sums over some time period to actual rates,
which can be interpreted by the user. This werk adjusts
the output units and metrics of these checks accordingly:
- aws_elb
- aws_elbv2_application.connections
- aws_elbv2_application.http_redirects
- aws_elbv2_application.statistics
- aws_elbv2_network.connections
- aws_elbv2_network.rst_packets
- aws_elbv2_network.statistics
- aws_elbv2_network.tls_handshakes
- aws_s3_requests.select_object
- aws_s3_requests.traffic_stats
Furthermore, for the following checks, the agent output has
been adjusted to be in line with what the user expects (e.g.
averages instead of sums for latencies, also partially connected
to werk #10962).
- aws_elbv2_application
- aws_elbv2_network
- aws_s3_requests.latency
Finally, a few bugs, such as integer instead of floating output,
have been fixed. Where necessary, rules and manpages have also been
updated.