* Add godot linter
Signed-off-by: George Robinson <george.robinson@grafana.com>
* Remove extra line from LICENSE
Signed-off-by: George Robinson <george.robinson@grafana.com>
---------
Signed-off-by: George Robinson <george.robinson@grafana.com>
Note that this does not stop showing classic metrics, for now
it is up to the scrape config to decide whether to keep those instead or
both.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
wait until `p2.Status()` returns because it blocks until we're ready - that way, we're guaranteed to know that the cluster size is 2.
Signed-off-by: gotjosh <josue.abreu@gmail.com>
* Add feature flag to enable discovery and use of public IPaddr for clustering.
Before this change, Alertmanager would refuse to startup if using a
advertise address binding to any address (0.0.0.0), and the host only
had an interface with a public IP address. After this change we feature
flag permitting the use of a discovered public address for cluster
gossiping.
Signed-off-by: Devin Trejo <dtrejo@palantir.com>
WaitReady is a blocking call and so should accept a Context in order to
be responsive to cancellation of the notification pipeline for any reason.
Signed-off-by: Steve Simpson <steve.simpson@grafana.com>
A Peer as defined by the `cluster` package represents the node in the
cluster. It is used in other packages to know the status of all of the
members or how long should we wait to know if a notification has already fired.
In Cortex, we'd like to implement a slightly different way of
clustering (using gRPC for communication and a
hash ring for node discovery).
This is a small change to support that by changing the consumer of other
packages to an interface.
Silences and Notification channels don't need an interface as they take
a `func([]byte) error` as a parameter.
Signed-off-by: gotjosh <josue@grafana.com>
- alertmanager_cluster_alive_messages_total, total number of alive
messages received.
- alertmanager_cluster_peer_info, a constant metric labeled by peer name.
- alertmanager_cluster_pings_seconds, histogram of latencies for ping
messages.
Signed-off-by: Simon Pasquier <spasquie@redhat.com>
This changes removes all usage of golang.org/x/net/context in the code
base. It also bumps a few dependencies for the same reason:
- github.com/gogo/protobuf
- go-openapi/*
Signed-off-by: Simon Pasquier <spasquie@redhat.com>
Adds a job which runs periodically and refreshes cluster.peer dns records.
The problem is that when you restart all of the alertmanager instances in an environment like Kubernetes, DNS may contain old alertmanager instance IPs, but on startup (when Join() happens) none of the new instance IPs. As at the start DNS is not empty resolvePeers waitIfEmpty=true, will return and "islands" of 1 alertmanager instances will form.
Signed-off-by: Povilas Versockas <p.versockas@gmail.com>
Alertmanager is exiting with a non-zero exit code if the initial cluster
join fails. This behavior could be not wanted because:
- As Alertmanager is a critical component with an at-least-once
guarantee, failing on joining the cluster is unnecessary as
Alertmanager still functions by itself.
- In an environment like Kubernetes discovering peers via DNS, peers
might roll out one-by-one, leaving the DNS entries unpopulated for the
first peer of a set. Failing on initial join prevents a roll-out.
Instead of failing on the initial join this patch only logs the failure.
The cluster can be later joined via the `handleReconnect`.
This is a regression introduced in PR #1456 [1].
[1] https://github.com/prometheus/alertmanager/pull/1456
Signed-off-by: Max Leonard Inden <IndenML@gmail.com>
* cluster: make sure we don't miss the first pushPull
During the join, memberlist initiates a pushPull to get initial data.
Unfortunately, at this point the nflog and silence listener have not
been registered yet, so the first data arrives only after one pushPull
cycle (1min by default !).
Signed-off-by: Corentin Chary <c.chary@criteo.com>