[openstack-dev] [Fuel] Removing old node logs

Andrey Danin adanin at mirantis.com
Thu Jun 26 10:38:36 UTC 2014


When nailgun remove a node it can gzip all the logs of this node to a
special file, like: /var/log/remote/archive/node-3-timestamp.tgz And
logrotate can keep these files for month, then delete them.

Master node health monitor is another big discussion.


On Thu, Jun 26, 2014 at 1:25 PM, Vladimir Kozhukalov <
vkozhukalov at mirantis.com> wrote:

> Making diagnostic snapshot for a particular environment is a good idea.
> But the issue is still there.
>
> We often have the situation when user actually doesn't care of old logs at
> all. He downloads ISO, installs it and tries various installation options
> (Ubuntu, Centos, HA, Ceph, etc.). Sooner or later his hard drive is full
> and he even can not make the diagnostic snapshot. Dealing with that stuff
> about taking care of available free space inside shotgun seems to be not a
> good idea. But we still need to address this. The easiest way to do that is
> to delete old log directories (logrotate or nailgun itself). In this case
> the issue at least will be rather seldom. But, of course, the right way is
> to have a kind of monitoring system on the master node and notify user when
> disk is full or launch a kind of cleaning task.
>
> Ok, right place where we should deal with that stuff about removing old
> logs is logrotate. Currently it just moves files like this
> /var/log/remote/old-node.example.com/some.log -> /var/log/remote/
> old-node.example.com/some.log.1.gz. But what it actually should do is to
> remove the whole directories which are related to nonexistent nodes, right?
>
>
>
>
>
> Vladimir Kozhukalov
>
>
> On Tue, Jun 24, 2014 at 9:19 PM, Andrey Danin <adanin at mirantis.com> wrote:
>
>> +1 to @Aleksandr
>>
>>
>> On Tue, Jun 24, 2014 at 8:32 PM, Aleksandr Didenko <adidenko at mirantis.com
>> > wrote:
>>
>>> Yes, of course, snapshot for all nodes at once (like currently) should
>>> also be available.
>>>
>>>
>>> On Tue, Jun 24, 2014 at 7:27 PM, Igor Kalnitsky <ikalnitsky at mirantis.com
>>> > wrote:
>>>
>>>> Hello,
>>>>
>>>> @Aleks, it's a good idea to make snapshot per environment, but I think
>>>> we can keep functionality to make snapshot for all nodes at once too.
>>>>
>>>> - Igor
>>>>
>>>>
>>>> On Tue, Jun 24, 2014 at 6:38 PM, Aleksandr Didenko <
>>>> adidenko at mirantis.com> wrote:
>>>>
>>>>> Yeah, I thought about diagnostic snapshot too. Maybe it would be
>>>>> better to implement per-environment diagnostic snapshots? I.e. add
>>>>> diagnostic snapshot generate/download buttons/links in the environment
>>>>> actions tab. Such snapshot would contain info/logs about Fuel master node
>>>>> and nodes assigned to the environment only.
>>>>>
>>>>>
>>>>> On Tue, Jun 24, 2014 at 6:27 PM, Igor Kalnitsky <
>>>>> ikalnitsky at mirantis.com> wrote:
>>>>>
>>>>>> Hi guys,
>>>>>>
>>>>>> What about our diagnostic snapshot?
>>>>>>
>>>>>> I mean we're going to make snapshot of entire /var/log and obviously
>>>>>> this old logs will be included in snapshot. Should we skip theem or
>>>>>> such situation is ok?
>>>>>>
>>>>>> - Igor
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Tue, Jun 24, 2014 at 5:57 PM, Aleksandr Didenko <
>>>>>> adidenko at mirantis.com> wrote:
>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> If user runs some experiments with creating/deleting clusters, then
>>>>>>> taking care of old logs is under user's responsibility, I suppose. Fuel
>>>>>>> configures log rotation with compression for remote logs, so old logs will
>>>>>>> be gzipped and will not take much space.
>>>>>>>
>>>>>>> In case of additional boolean parameter, the default value should be
>>>>>>> "0-don't touch old logs".
>>>>>>>
>>>>>>> --
>>>>>>> Regards,
>>>>>>> Alex
>>>>>>>
>>>>>>>
>>>>>>> On Tue, Jun 24, 2014 at 4:07 PM, Vladimir Kozhukalov <
>>>>>>> vkozhukalov at mirantis.com> wrote:
>>>>>>>
>>>>>>>> Guys,
>>>>>>>>
>>>>>>>> What do you think of removing node logs on master node right after
>>>>>>>> removing node from cluster?
>>>>>>>>
>>>>>>>> The issue is when user do experiments he creates and deletes
>>>>>>>> clusters and old unused directories remain and take disk space. On the
>>>>>>>> other hand, it is not so hard to imaging the situation when user would like
>>>>>>>> to be able to take a look in old logs.
>>>>>>>>
>>>>>>>> My suggestion here is to add a boolean parameter into settings
>>>>>>>> which will manage this piece of logic (1-remove old logs, 0-don't touch old
>>>>>>>> logs).
>>>>>>>>
>>>>>>>> Thanks for your opinions.
>>>>>>>>
>>>>>>>> Vladimir Kozhukalov
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> OpenStack-dev mailing list
>>>>>>>> OpenStack-dev at lists.openstack.org
>>>>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> OpenStack-dev mailing list
>>>>>>> OpenStack-dev at lists.openstack.org
>>>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>>>>
>>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> OpenStack-dev mailing list
>>>>>> OpenStack-dev at lists.openstack.org
>>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OpenStack-dev mailing list
>>>>> OpenStack-dev at lists.openstack.org
>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> OpenStack-dev mailing list
>>>> OpenStack-dev at lists.openstack.org
>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>
>>>>
>>>
>>> _______________________________________________
>>> OpenStack-dev mailing list
>>> OpenStack-dev at lists.openstack.org
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>
>>>
>>
>>
>> --
>> Andrey Danin
>> adanin at mirantis.com
>> skype: gcon.monolake
>>
>> _______________________________________________
>> OpenStack-dev mailing list
>> OpenStack-dev at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>


-- 
Andrey Danin
adanin at mirantis.com
skype: gcon.monolake
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140626/3d194021/attachment.html>


More information about the OpenStack-dev mailing list