Just checking in

15 views
Skip to first unread message

Uwe Kubosch

unread,
Jan 30, 2011, 7:14:24 AM1/30/11
to zfs-...@googlegroups.com
Hi all!

How are you all doing?

It has been a while since I released a zfs-fuse package for Fedora/EPEL. The last was zfs-fuse-0.6.9-7.20100709git

From recent messages it seems 0.7.0 is not far away. If you would like any changes in the packaging for Fedora/EPEL, please let me know.

--
With kind regards
Uwe Kubosch
Kubosch Consulting
u...@kubosch.no
http://kubosch.no/

sgheeren

unread,
Jan 30, 2011, 9:12:39 AM1/30/11
to zfs-...@googlegroups.com
Yes, Uwe, just the otherday this was brought to my attention. I believe
your init script sets the oom adjust for zfs-fuse as well?

http://zfs-fuse.net/issues/116

Uwe Kubosch

unread,
Jan 30, 2011, 5:30:45 PM1/30/11
to zfs-...@googlegroups.com

On 2011-01-30, at 15:12, sgheeren wrote:
> On 01/30/2011 01:14 PM, Uwe Kubosch wrote:
>> It has been a while since I released a zfs-fuse package for Fedora/EPEL. The last was zfs-fuse-0.6.9-7.20100709git
>> From recent messages it seems 0.7.0 is not far away. If you would like any changes in the packaging for Fedora/EPEL, please let me know.
> Yes, Uwe, just the otherday this was brought to my attention. I believe
> your init script sets the oom adjust for zfs-fuse as well?
>
> http://zfs-fuse.net/issues/116


So, I changed

echo -17 > "/proc/$PID/oom_adj"

to

echo -17 > "/proc/$PID/oom_score_adj"

Will that suffice?

At the next release I will attempt to use the init script in the zfs-fuse tarball instead of a Fedora/EPEL specific script. Most likely I will request changes :)

sgheeren

unread,
Jan 30, 2011, 6:21:53 PM1/30/11
to zfs-...@googlegroups.com
On 01/30/2011 11:30 PM, Uwe Kubosch wrote:
> On 2011-01-30, at 15:12, sgheeren wrote:
>> On 01/30/2011 01:14 PM, Uwe Kubosch wrote:
>>> It has been a while since I released a zfs-fuse package for Fedora/EPEL. The last was zfs-fuse-0.6.9-7.20100709git
>>> From recent messages it seems 0.7.0 is not far away. If you would like any changes in the packaging for Fedora/EPEL, please let me know.
>> Yes, Uwe, just the otherday this was brought to my attention. I believe
>> your init script sets the oom adjust for zfs-fuse as well?
>>
>> http://zfs-fuse.net/issues/116
>
> So, I changed
>
> echo -17 > "/proc/$PID/oom_adj"
>
> to
>
> echo -17 > "/proc/$PID/oom_score_adj"
>
> Will that suffice?
>
I'm not sure, at least the text in the commit said that there should be
some kind of non-linear translation function from old to new

> At the next release I will attempt to use the init script in the zfs-fuse tarball instead of a Fedora/EPEL specific script. Most likely I will request changes :)
Very good, I'm happy to oblige!

Fajar A. Nugraha

unread,
Jan 30, 2011, 9:08:52 PM1/30/11
to zfs-...@googlegroups.com
On Mon, Jan 31, 2011 at 6:21 AM, sgheeren <sghe...@hotmail.com> wrote:
> On 01/30/2011 11:30 PM, Uwe Kubosch wrote:
>> On 2011-01-30, at 15:12, sgheeren wrote:
>>> On 01/30/2011 01:14 PM, Uwe Kubosch wrote:
>>>> It has been a while since I released a zfs-fuse package for Fedora/EPEL.  The last was zfs-fuse-0.6.9-7.20100709git
>>>> From recent messages it seems 0.7.0 is not far away.  If you would like any changes in the packaging for Fedora/EPEL, please let me know.
>>> Yes, Uwe, just the otherday this was brought to my attention. I believe
>>> your init script sets the oom adjust for zfs-fuse as well?
>>>
>>> http://zfs-fuse.net/issues/116
>>
>> So, I changed
>>
>>     echo -17 > "/proc/$PID/oom_adj"
>>
>> to
>>
>>     echo -17 > "/proc/$PID/oom_score_adj"
>>
>> Will that suffice?
>>
> I'm not sure, at least the text in the commit said that there should be
> some kind of non-linear translation function from old to new

yup,

something like
if [ -f "/proc/$PID/oom_score_adj" ]; then
echo -17 > "/proc/$PID/oom_score_adj"
else
echo -17 > "/proc/$PID/oom_adj"
fi


>> At the next release I will attempt to use the init script in the zfs-fuse tarball instead of a Fedora/EPEL specific script.  Most likely I will request changes :)
> Very good, I'm happy to oblige!

What is the latest pool version that zfs-fuse supports (whichever
branch that will be 0.7)? The last EPEL package was on 22, while
native zfs is at 28.

--
Fajar

Uwe Kubosch

unread,
Jan 31, 2011, 4:22:08 AM1/31/11
to zfs-...@googlegroups.com

On 2011-01-31, at 03:08, Fajar A. Nugraha wrote:
> On Mon, Jan 31, 2011 at 6:21 AM, sgheeren <sghe...@hotmail.com> wrote:
>> On 01/30/2011 11:30 PM, Uwe Kubosch wrote:
>>> On 2011-01-30, at 15:12, sgheeren wrote:
>>>> On 01/30/2011 01:14 PM, Uwe Kubosch wrote:
>>>>> It has been a while since I released a zfs-fuse package for Fedora/EPEL. The last was zfs-fuse-0.6.9-7.20100709git
>>>>> From recent messages it seems 0.7.0 is not far away. If you would like any changes in the packaging for Fedora/EPEL, please let me know.
>>>> Yes, Uwe, just the otherday this was brought to my attention. I believe
>>>> your init script sets the oom adjust for zfs-fuse as well?
>>>> http://zfs-fuse.net/issues/116
>>>
>>> So, I changed
>>> echo -17 > "/proc/$PID/oom_adj"
>>> to
>>> echo -17 > "/proc/$PID/oom_score_adj"
>>> Will that suffice?
>> I'm not sure, at least the text in the commit said that there should be
>> some kind of non-linear translation function from old to new
> something like
> if [ -f "/proc/$PID/oom_score_adj" ]; then
> echo -17 > "/proc/$PID/oom_score_adj"
> else
> echo -17 > "/proc/$PID/oom_adj"
> fi


Added to init script. Will appear in the next release.

Seth Heeren

unread,
Jan 31, 2011, 6:23:19 AM1/31/11
to zfs-fuse
I still don't think that is accurate. Like I said there is some kind
of translation function. From the web:

oom-score = oom-base-score x 2^oom_adj

new tuning parameter mean)

oom-score = oom-base-score + oom_score_adj / (totalram +
totalswap)

Also, from http://lwn.net/Articles/396552/

Instead, a new tunable, /proc/pid/oom_score_adj, is added that
ranges from -1000 to +1000

I really don't have the time to find out the details here, but bluntly
replacing the filespecification is not going to cut it (why would
there even have been a change in the first place :))

Daniel Brooks

unread,
Feb 3, 2011, 11:27:40 AM2/3/11
to zfs-...@googlegroups.com

Uwe, do you keep a repository for this rpm? I'd like to take a look at it if I may.

Uwe Kubosch

unread,
Feb 3, 2011, 6:51:38 PM2/3/11
to zfs-...@googlegroups.com
On 2011-02-03, at 17:27, Daniel Brooks wrote:

Uwe, do you keep a repository for this rpm? I'd like to take a look at it if I may.

sgheeren

unread,
Mar 9, 2011, 8:49:09 AM3/9/11
to zfs-...@googlegroups.com
Uwe, any news?

I'm currently in the process of releasing the 0.7.0 version (grab it
ahead of the fanfare here:

http://gitweb.zfs-fuse.net/?p=sehe;a=shortlog;h=refs/heads/master (of
course, the official's maint branch will be recommended for packaging,
but this branch is already correct for the release)

or in my ppa:
https://launchpad.net/~bugs-sehe/+archive/zfs-fuse/+packages
<https://launchpad.net/%7Ebugs-sehe/+archive/zfs-fuse/+packages>

Cheers,
Seth

Uwe Kubosch

unread,
Mar 11, 2011, 6:08:10 PM3/11/11
to zfs-...@googlegroups.com

On 2011-03-09, at 14:49, sgheeren wrote:

> Uwe, any news?

Huh? Did you wait for something from me :)

> I'm currently in the process of releasing the 0.7.0 version (grab it
> ahead of the fanfare here:

A bit pressed on time. Will not be able to look at it any time soon , I think. I can give access to the repository and packaging if someone want to speed things up.

> --
> To post to this group, send email to zfs-...@googlegroups.com
> To visit our Web site, click on http://zfs-fuse.net/

sgheeren

unread,
Mar 11, 2011, 7:08:21 PM3/11/11
to zfs-...@googlegroups.com
On 03/12/2011 12:08 AM, Uwe Kubosch wrote:
> On 2011-03-09, at 14:49, sgheeren wrote:
>
>> Uwe, any news?
> Huh? Did you wait for something from me :)
I'm not waiting. (I don't need RPMs...)
I hope you _did_ notice that I question the outcome of the OOM
discussion as it was 'hanging'. This is why I bumped this particular
thread :)

Once I get time, I might do the maths on it and you can just take it
from my upstart script. Meanwhile, I trust the 'legacy' OOM interface

>> I'm currently in the process of releasing the 0.7.0 version (grab it
>> ahead of the fanfare here:
> A bit pressed on time. Will not be able to look at it any time soon , I think. I can give access to the repository and packaging if someone want to speed things up.

Likewise here, no problem. In case anyone really craves RPMs in EPEL, I
think we will hear them here,

Thanks,
Seth

Uwe Kubosch

unread,
Mar 12, 2011, 4:39:27 PM3/12/11
to zfs-...@googlegroups.com

On 2011-03-12, at 01:08, sgheeren wrote:

> On 03/12/2011 12:08 AM, Uwe Kubosch wrote:
>> On 2011-03-09, at 14:49, sgheeren wrote:
>>
>>> Uwe, any news?
>> Huh? Did you wait for something from me :)
> I'm not waiting. (I don't need RPMs...)
> I hope you _did_ notice that I question the outcome of the OOM
> discussion as it was 'hanging'. This is why I bumped this particular
> thread :)

Yes, I was waiting for someone to do the math work :)

> Once I get time, I might do the maths on it and you can just take it
> from my upstart script. Meanwhile, I trust the 'legacy' OOM interface

Good.

>>> I'm currently in the process of releasing the 0.7.0 version (grab it
>>> ahead of the fanfare here:
>> A bit pressed on time. Will not be able to look at it any time soon , I think. I can give access to the repository and packaging if someone want to speed things up.
> Likewise here, no problem. In case anyone really craves RPMs in EPEL, I
> think we will hear them here,

Fine by me. Just starting a new project at work, and it takes all my time.

sgheeren

unread,
Mar 12, 2011, 4:45:33 PM3/12/11
to zfs-...@googlegroups.com
On 03/12/2011 10:39 PM, Uwe Kubosch wrote:
> Fine by me. Just starting a new project at work, and it takes all my time.
Don't you just hate it when work get's in the way of other work :)
Reply all
Reply to author
Forward
0 new messages