Re: GlusterFS and Kafka

[Christopher Schmidt <fakod666@xxxxxxxxx>]

I was stupid enough to copy an additional newline from email. So, sorry for the noise.

Works so far.

Thanks for getting that solved, best Chris

Raghavendra Talur <rtalur@xxxxxxxxxx> schrieb am Mo., 29. Mai 2017 um 13:18 Uhr:

On 29-May-2017 3:49 PM, "Christopher Schmidt" <fakod666@xxxxxxxxx> wrote:

Hi Raghavendra Talur,

this does not work for me. Most certainly because I forgot something.

So just put the file in the folder, make it executable and create a volume? Thats all?

If I am doing this, there is no /var/lib/glusterd/hooks/1/create/post/log file and the Performance Translator is still on.

and idea?

Most certainly selinux. I think you will have to set context on the file to same as others in hooks dir. 

You can test temporarily by

setenforce  0

Raghavendra Talur <rtalur@xxxxxxxxxx> schrieb am Fr., 26. Mai 2017 um 07:34 Uhr:

On Thu, May 25, 2017 at 8:39 PM, Joe Julian <joe@xxxxxxxxxxxxxxxx> wrote:
> Maybe hooks?

Yes, we were thinking of the same :)

Christopher,
Gluster has hook-scripts facility that admins can write and set those
to be run on certain events in Gluster. We have a event for volume
creation.
Here are the steps for using hook scripts.

1. deploy the gluster pods and create a cluster as you have already done.
2. on the kubernetes nodes that are running gluster pods(make sure
they are running now, because we want to write into the bind mount),
create a new file in location /var/lib/glusterd/hooks/1/create/post/
3. name of the fie could be S29disable-perf.sh , important part being
that the number should have capital S as first letter.
4. I tried out a sample script with content as below

```
#!/bin/bash



PROGNAME="Sdisable-perf"
OPTSPEC="volname:,gd-workdir:"
VOL=
CONFIGFILE=
LOGFILEBASE=
PIDDIR=
GLUSTERD_WORKDIR=

function parse_args () {
        ARGS=$(getopt -l $OPTSPEC  -name $PROGNAME $@)
        eval set -- "$ARGS"

        while true; do
            case $1 in
                --volname)
                    shift
                    VOL=$1
                    ;;
                --gd-workdir)
                    shift
                    GLUSTERD_WORKDIR=$1
                    ;;
                *)
                    shift
                    break
                    ;;
            esac
            shift
        done
}

function disable_perf_xlators () {
        volname=$1
        gluster volume set $volname performance.write-behind off
        echo "executed and return is $?" >>
/var/lib/glusterd/hooks/1/create/post/log
}

echo "starting" >> /var/lib/glusterd/hooks/1/create/post/log
parse_args $@
disable_perf_xlators $VOL
```
5. set execute permissions on the file

I tried this out and it worked for me. Let us know if that helps!

Thanks,
Raghavendra Talur



>
>
> On May 25, 2017 6:48:04 AM PDT, Christopher Schmidt <fakod666@xxxxxxxxx>
> wrote:
>>
>> Hi Humble,
>>
>> thanks for that, it is really appreciated.
>>
>> In the meanwhile, using K8s 1.5, what can I do to disable the performance
>> translator that doesn't work with Kafka? Maybe something while generating
>> the Glusterfs container for Kubernetes?
>>
>> Best Christopher
>>
>> Humble Chirammal <hchiramm@xxxxxxxxxx> schrieb am Do., 25. Mai 2017,
>> 09:36:
>>>
>>> On Thu, May 25, 2017 at 12:57 PM, Raghavendra Talur <rtalur@xxxxxxxxxx>
>>> wrote:
>>>>
>>>> On Thu, May 25, 2017 at 11:21 AM, Christopher Schmidt
>>>> <fakod666@xxxxxxxxx> wrote:
>>>> > So this change of the Gluster Volume Plugin will make it into K8s 1.7
>>>> > or
>>>> > 1.8. Unfortunately too late for me.
>>>> >
>>>> > Does anyone know how to disable performance translators by default?
>>>>
>>>> Humble,
>>>>
>>>> Do you know of any way Christopher can proceed here?
>>>
>>>
>>> I am trying to get it in 1.7 branch, will provide an update here as soon
>>> as its available.
>>>>
>>>>
>>>> >
>>>> >
>>>> > Raghavendra Talur <rtalur@xxxxxxxxxx> schrieb am Mi., 24. Mai 2017,
>>>> > 19:30:
>>>> >>
>>>> >> On Wed, May 24, 2017 at 4:08 PM, Christopher Schmidt
>>>> >> <fakod666@xxxxxxxxx>
>>>> >> wrote:
>>>> >> >
>>>> >> >
>>>> >> > Vijay Bellur <vbellur@xxxxxxxxxx> schrieb am Mi., 24. Mai 2017 um
>>>> >> > 05:53
>>>> >> > Uhr:
>>>> >> >>
>>>> >> >> On Tue, May 23, 2017 at 1:39 AM, Christopher Schmidt
>>>> >> >> <fakod666@xxxxxxxxx>
>>>> >> >> wrote:
>>>> >> >>>
>>>> >> >>> OK, seems that this works now.
>>>> >> >>>
>>>> >> >>> A couple of questions:
>>>> >> >>> - What do you think, are all these options necessary for Kafka?
>>>> >> >>
>>>> >> >>
>>>> >> >> I am not entirely certain what subset of options will make it work
>>>> >> >> as I
>>>> >> >> do
>>>> >> >> not understand the nature of failure with  Kafka and the default
>>>> >> >> gluster
>>>> >> >> configuration. It certainly needs further analysis to identify the
>>>> >> >> list
>>>> >> >> of
>>>> >> >> options necessary. Would it be possible for you to enable one
>>>> >> >> option
>>>> >> >> after
>>>> >> >> the other and determine the configuration that ?
>>>> >> >>
>>>> >> >>
>>>> >> >>>
>>>> >> >>> - You wrote that there have to be kind of application profiles.
>>>> >> >>> So to
>>>> >> >>> find out, which set of options work is currently a matter of
>>>> >> >>> testing
>>>> >> >>> (and
>>>> >> >>> hope)? Or are there any experiences for MongoDB / ProstgreSQL /
>>>> >> >>> Zookeeper
>>>> >> >>> etc.?
>>>> >> >>
>>>> >> >>
>>>> >> >> Application profiles are work in progress. We have a few that are
>>>> >> >> focused
>>>> >> >> on use cases like VM storage, block storage etc. at the moment.
>>>> >> >>
>>>> >> >>>
>>>> >> >>> - I am using Heketi and Dynamik Storage Provisioning together
>>>> >> >>> with
>>>> >> >>> Kubernetes. Can I set this volume options somehow by default or
>>>> >> >>> by
>>>> >> >>> volume
>>>> >> >>> plugin?
>>>> >> >>
>>>> >> >>
>>>> >> >>
>>>> >> >> Adding Raghavendra and Michael to help address this query.
>>>> >> >
>>>> >> >
>>>> >> > For me it would be sufficient to disable some (or all) translators,
>>>> >> > for
>>>> >> > all
>>>> >> > volumes that'll be created, somewhere here:
>>>> >> > https://github.com/gluster/gluster-containers/tree/master/CentOS
>>>> >> > This is the container used by the GlusterFS DaemonSet for
>>>> >> > Kubernetes.
>>>> >>
>>>> >> Work is in progress to give such option at volume plugin level. We
>>>> >> currently have a patch[1] in review for Heketi that allows users to
>>>> >> set Gluster options using heketi-cli instead of going into a Gluster
>>>> >> pod. Once this is in, we can add options in storage-class of
>>>> >> Kubernetes that pass down Gluster options for every volume created in
>>>> >> that storage-class.
>>>> >>
>>>> >> [1] https://github.com/heketi/heketi/pull/751
>>>> >>
>>>> >> Thanks,
>>>> >> Raghavendra Talur
>>>> >>
>>>> >> >
>>>> >> >>
>>>> >> >>
>>>> >> >> -Vijay
>>>> >> >>
>>>> >> >>
>>>> >> >>
>>>> >> >>>
>>>> >> >>>
>>>> >> >>> Thanks for you help... really appreciated.. Christopher
>>>> >> >>>
>>>> >> >>> Vijay Bellur <vbellur@xxxxxxxxxx> schrieb am Mo., 22. Mai 2017 um
>>>> >> >>> 16:41
>>>> >> >>> Uhr:
>>>> >> >>>>
>>>> >> >>>> Looks like a problem with caching. Can you please try by
>>>> >> >>>> disabling
>>>> >> >>>> all
>>>> >> >>>> performance translators? The following configuration commands
>>>> >> >>>> would
>>>> >> >>>> disable
>>>> >> >>>> performance translators in the gluster client stack:
>>>> >> >>>>
>>>> >> >>>> gluster volume set <volname> performance.quick-read off
>>>> >> >>>> gluster volume set <volname> performance.io-cache off
>>>> >> >>>> gluster volume set <volname> performance.write-behind off
>>>> >> >>>> gluster volume set <volname> performance.stat-prefetch off
>>>> >> >>>> gluster volume set <volname> performance.read-ahead off
>>>> >> >>>> gluster volume set <volname> performance.readdir-ahead off
>>>> >> >>>> gluster volume set <volname> performance.open-behind off
>>>> >> >>>> gluster volume set <volname> performance.client-io-threads off
>>>> >> >>>>
>>>> >> >>>> Thanks,
>>>> >> >>>> Vijay
>>>> >> >>>>
>>>> >> >>>>
>>>> >> >>>>
>>>> >> >>>> On Mon, May 22, 2017 at 9:46 AM, Christopher Schmidt
>>>> >> >>>> <fakod666@xxxxxxxxx> wrote:
>>>> >> >>>>>
>>>> >> >>>>> Hi all,
>>>> >> >>>>>
>>>> >> >>>>> has anyone ever successfully deployed a Kafka (Cluster) on
>>>> >> >>>>> GlusterFS
>>>> >> >>>>> volumes?
>>>> >> >>>>>
>>>> >> >>>>> I my case it's a Kafka Kubernetes-StatefulSet and a Heketi
>>>> >> >>>>> GlusterFS.
>>>> >> >>>>> Needless to say that I am getting a lot of filesystem related
>>>> >> >>>>> exceptions like this one:
>>>> >> >>>>>
>>>> >> >>>>> Failed to read `log header` from file channel
>>>> >> >>>>> `sun.nio.ch.FileChannelImpl@67afa54a`. Expected to read 12
>>>> >> >>>>> bytes,
>>>> >> >>>>> but
>>>> >> >>>>> reached end of file after reading 0 bytes. Started read from
>>>> >> >>>>> position
>>>> >> >>>>> 123065680.
>>>> >> >>>>>
>>>> >> >>>>> I limited the amount of exceptions with the
>>>> >> >>>>> log.flush.interval.messages=1 option, but not all...
>>>> >> >>>>>
>>>> >> >>>>> best Christopher
>>>> >> >>>>>
>>>> >> >>>>>
>>>> >> >>>>> _______________________________________________
>>>> >> >>>>> Gluster-users mailing list
>>>> >> >>>>> Gluster-users@xxxxxxxxxxx
>>>> >> >>>>> http://lists.gluster.org/mailman/listinfo/gluster-users
>>>> >> >>>>
>>>> >> >>>>
>>>> >> >
>>>
>>>
>>>
>>>
>>> --
>>> Cheers,
>>> Humble
>>>
>>> Sr.Software Engineer - Red Hat Storage Engineering
>>> website: http://humblec.com
>
>
> --
> Sent from my Android device with K-9 Mail. Please excuse my brevity.
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users@xxxxxxxxxxx
> http://lists.gluster.org/mailman/listinfo/gluster-users

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users