Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Trigger alert for vol and brick failures #287

Closed
wants to merge 3 commits into from
Closed

Trigger alert for vol and brick failures #287

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f2fb1b6
Change the structure of gluster bricks
nnDarshan Jun 2, 2017
f428567
Raise alerts for volume and brick status change
nnDarshan Jun 6, 2017
4b6a5b8
Merge branch 'develop' into TriggerAlertForVolAndBrickFailures
r0h4n Jun 6, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
96 changes: 96 additions & 0 deletions tendrl/gluster_integration/sds_sync/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
import json
import os
import re
import socket
import subprocess

from tendrl.gluster_integration.sds_sync import brick_utilization
Expand All @@ -12,6 +13,8 @@

from tendrl.commons import sds_sync
from tendrl.gluster_integration import ini2json
from tendrl.commons.utils.time_utils import now as tendrl_now



class GlusterIntegrationSdsSyncStateThread(sds_sync.SdsSyncThread):
Expand All @@ -20,6 +23,37 @@ def __init__(self):
super(GlusterIntegrationSdsSyncStateThread, self).__init__()
self._complete = gevent.event.Event()

def _emit_event(self, resource, curr_value, msg):
alert = {}
alert['source'] = 'tendrl-gluster-integration'
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

use NS.publisher_id

alert['pid'] = os.getpid()
alert['timestamp'] = tendrl_now().isoformat()
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should be time_stamp as per latest testing with alerting module.

alert['alert_type'] = 'status'
severity = "info"
if curr_value == "Stopped":
severity = "critical"
alert['severity'] = severity
alert['resource'] = resource
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Resource name should be something like volume_status and the same should be populated as handles field of a handler which needs to added under tendrl-alerting module as handlers/cluster/volume_status_handler.py

You can refer Tendrl/notifier#78 for more details.

If required a separate handler to be written for brick status changes.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The resource should be the name of object as defined in Tendrl definitions i.e. Volume and the other part of resource should be the attribute on which the alert is triggered

Volume.status

Copy link
Contributor

@anmolbabu anmolbabu Jun 7, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@r0h4n The alerting module expects it as volume_status or volume_utilization so basically its the combination of <entity_type>_<alert_type> where alert_type is either status or utilization...
This avoids the alerting module having to know the object definition from the definitions of every integration module..

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The alerting module does need to know about the object definition and attributes. Tendrl will be generating alerts on specific attributes of a Tendrl object.

But if you are expecting a underscore between the entity/object and alert/attribute we can continue with below scheme

_

works?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes .. volume_status works with alerting as it stands today...

alert['current_value'] = curr_value
alert['tags'] = dict(
message=msg,
cluster_id=NS.tendrl_context.cluster_id,
cluster_name=NS.tendrl_context.cluster_name,
sds_name=NS.tendrl_context.sds_name,
fqdn=socket.getfqdn()
)
alert['node_id'] = NS.node_context.node_id
if not NS.node_context.node_id:
return
Event(
Message(
"notice",
"alerting",
{'message': json.dumps(alert)},
node_id=NS.node_context.node_id
Copy link
Contributor

@anmolbabu anmolbabu Jun 7, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This(node_id) is not required it will be automatically taken by Message class..

)
)

def _run(self):
Event(
Message(
Expand Down Expand Up @@ -82,8 +116,35 @@ def _run(self):
if "Volumes" in raw_data:
index = 1
volumes = raw_data['Volumes']
node_context = NS.node_context.load()
tag_list = list(node_context.tags)
while True:
try:
# Raise alerts for volume state change.
cluster_provisioner = "provisioner/%s" % NS.tendrl_context.integration_id
if cluster_provisioner in tag_list:
try:
stored_volume_status = NS._int.client.read(
"clusters/%s/Volumes/%s/status" % (
NS.tendrl_context.integration_id,
volumes['volume%s.id' % index]
)
).value
current_status = volumes['volume%s.status' % index]
if current_status != stored_volume_status:
msg = "Status of volume: %s changed from %s to %s" % (
volumes['volume%s.name' % index],
stored_volume_status,
current_status
)
self._emit_event(
volumes['volume%s.name' % index],
current_status,
msg)

except etcd.EtcdKeyNotFound:
pass

volume = NS.gluster.objects.Volume(
vol_id=volumes[
'volume%s.id' % index
Expand Down Expand Up @@ -187,6 +248,7 @@ def _run(self):
hostname not in network_ip):
b_index += 1
continue

sub_vol_size = (int(volumes['volume%s.brickcount' % index])) / int(
volumes[
'volume%s.subvol_count' % index
Expand All @@ -198,6 +260,40 @@ def _run(self):
)
].split(":")[-1].replace("/","_")

# Raise alerts if the brick path changes
try:
stored_brick_status = NS._int.client.read(
"clusters/%s/Bricks/all/%s/status" % (
NS.tendrl_context.integration_id,
brick_name
)
).value
current_status = volumes.get(
'volume%s.brick%s.status' % (
index,
b_index
)
)
if current_status != stored_brick_status:
msg = "Status of brick: %s under volume %s changed from %s to %s" % (
volumes[
'volume%s.brick%s.path' % (
index, b_index
)
],
volumes['volume%s.name' % index],
stored_brick_status,
current_status
)
self._emit_event(
volumes['volume%s.name' % index],
Copy link
Contributor

@anmolbabu anmolbabu Jun 7, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

for cases like brick where info about its volume and cluster both are important, have resource type as brick_status and have an additional parameter under tags called the plugin_instance and have its value in accordance with https://github.com/Tendrl/node-monitoring/blob/develop/tendrl/node_monitoring/plugins/tendrl_glusterfs_brick_utilization.py#L286

Now, the reason for not having additional fields like vol_name, brick_path extra under the dict tags is , that collectd does not allow us to have custom additional fields in tags as the tags attribute for collectd generated alerts come directly from collectd based on how the plugin is configured(an example is the above link) and only a few reserved fields can be played around with which leaves plugin_instance as the best attribute choice left...
So if we decide to have custom fields in our(tendrl generated) alerts unless absolutely necessary will render the alerting module validations and alert clearing logic less generic...

current_status,
msg
)

except etcd.EtcdKeyNotFound:
pass

vol_brick_path = "clusters/%s/Volumes/%s/Bricks/subvolume%s/%s" % (
NS.tendrl_context.integration_id,
volumes['volume%s.id' % index],
Expand Down